section5.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "toc_visible": true,
      "authorship_tag": "ABX9TyNJygowbLiPMbexdKvhrOzD",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "gpuClass": "standard"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/GustaFTW/TensorFlow-ZTM/blob/main/section5.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Transfer Learning with TensorFlow Part 2: Fine-Tuning\n",
        "\n",
        "In the preivous notebook., we covereded transer learning feature extraction, now it's time to learn about a new kind of transfer learning: fine-tuning"
      ],
      "metadata": {
        "id": "LANrCaZ5pvTI"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Check if w're using a GPU\n",
        "!nvidia-smi"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "s65ZOHudszBn",
        "outputId": "044dc37e-730c-4f61-c4d9-feadc4663e20"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Fri Jan  6 16:24:29 2023       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 460.32.03    Driver Version: 460.32.03    CUDA Version: 11.2     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|                               |                      |               MIG M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla T4            Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   64C    P0    31W /  70W |      0MiB / 15109MiB |      0%      Default |\n",
            "|                               |                      |                  N/A |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                                  |\n",
            "|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |\n",
            "|        ID   ID                                                   Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Creating helper function\n",
        "\n",
        "In previous notebooks, we've create a bunck of helper functions, now we could rewrite them all, however, this is tedious.\n",
        "\n",
        "So, it's a good idea to put functions you'll want to use again in a script you can download and import into your notebooks (or elsewhere).\n",
        "We've done this for some of the functions we've used previously here: https://raw.githubusercontent.com/mrdbourke/tensorflow-deep-learning/main/extras/helper_functions.py"
      ],
      "metadata": {
        "id": "8tbZgFhDs1zJ"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!wget https://raw.githubusercontent.com/mrdbourke/tensorflow-deep-learning/main/extras/helper_functions.py"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JQ4yy54_tA6E",
        "outputId": "79a90f71-cae6-4500-98a4-88f8a911823f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "--2023-01-06 16:24:29--  https://raw.githubusercontent.com/mrdbourke/tensorflow-deep-learning/main/extras/helper_functions.py\n",
            "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 185.199.111.133, 185.199.109.133, 185.199.108.133, ...\n",
            "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|185.199.111.133|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 10246 (10K) [text/plain]\n",
            "Saving to: ‘helper_functions.py’\n",
            "\n",
            "\rhelper_functions.py   0%[                    ]       0  --.-KB/s               \rhelper_functions.py 100%[===================>]  10.01K  --.-KB/s    in 0s      \n",
            "\n",
            "2023-01-06 16:24:29 (107 MB/s) - ‘helper_functions.py’ saved [10246/10246]\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Import helper function we're going to use in this notebook\n",
        "from helper_functions import create_tensorboard_callback, plot_loss_curves, unzip_data, walk_through_dir"
      ],
      "metadata": {
        "id": "VjJ5NWw5tWU2"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "> **Note:** If you're running this notebook in Google Colab, when it times out Colab will delete `helper_function.py`, so you'll have to redownload it if you want access yo your helper functions."
      ],
      "metadata": {
        "id": "jbATK2Obt814"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Let's get some data\n",
        "\n",
        "This time we're going to see how we can use the pretrained models within tf.keras.applications and apply them to our own problem (recognizing images of food).\n",
        "\n",
        "link: https://tensorflow.org/api_docs/python/tf/keras/applications"
      ],
      "metadata": {
        "id": "jqQJDuE4uQ-M"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Get 10% of training data of 10 classes of Food101\n",
        "!wget https://storage.googleapis.com/ztm_tf_course/food_vision/10_food_classes_10_percent.zip\n",
        "unzip_data(\"10_food_classes_10_percent.zip\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ifrs8WSFuYiY",
        "outputId": "46290f30-ac1a-45aa-f34d-68d0f88a72bd"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "--2023-01-06 16:24:33--  https://storage.googleapis.com/ztm_tf_course/food_vision/10_food_classes_10_percent.zip\n",
            "Resolving storage.googleapis.com (storage.googleapis.com)... 173.194.214.128, 173.194.215.128, 173.194.216.128, ...\n",
            "Connecting to storage.googleapis.com (storage.googleapis.com)|173.194.214.128|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 168546183 (161M) [application/zip]\n",
            "Saving to: ‘10_food_classes_10_percent.zip’\n",
            "\n",
            "10_food_classes_10_ 100%[===================>] 160.74M   206MB/s    in 0.8s    \n",
            "\n",
            "2023-01-06 16:24:34 (206 MB/s) - ‘10_food_classes_10_percent.zip’ saved [168546183/168546183]\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Check out how many images and subdirectories are in our dataset\n",
        "walk_through_dir(\"10_food_classes_10_percent\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "UYtyfOLNvGRG",
        "outputId": "4e30f4d7-88f7-46aa-a54e-321815abacc1"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "There are 2 directories and 0 images in '10_food_classes_10_percent'.\n",
            "There are 10 directories and 0 images in '10_food_classes_10_percent/test'.\n",
            "There are 0 directories and 250 images in '10_food_classes_10_percent/test/chicken_wings'.\n",
            "There are 0 directories and 250 images in '10_food_classes_10_percent/test/steak'.\n",
            "There are 0 directories and 250 images in '10_food_classes_10_percent/test/hamburger'.\n",
            "There are 0 directories and 250 images in '10_food_classes_10_percent/test/sushi'.\n",
            "There are 0 directories and 250 images in '10_food_classes_10_percent/test/ice_cream'.\n",
            "There are 0 directories and 250 images in '10_food_classes_10_percent/test/grilled_salmon'.\n",
            "There are 0 directories and 250 images in '10_food_classes_10_percent/test/chicken_curry'.\n",
            "There are 0 directories and 250 images in '10_food_classes_10_percent/test/ramen'.\n",
            "There are 0 directories and 250 images in '10_food_classes_10_percent/test/pizza'.\n",
            "There are 0 directories and 250 images in '10_food_classes_10_percent/test/fried_rice'.\n",
            "There are 10 directories and 0 images in '10_food_classes_10_percent/train'.\n",
            "There are 0 directories and 75 images in '10_food_classes_10_percent/train/chicken_wings'.\n",
            "There are 0 directories and 75 images in '10_food_classes_10_percent/train/steak'.\n",
            "There are 0 directories and 75 images in '10_food_classes_10_percent/train/hamburger'.\n",
            "There are 0 directories and 75 images in '10_food_classes_10_percent/train/sushi'.\n",
            "There are 0 directories and 75 images in '10_food_classes_10_percent/train/ice_cream'.\n",
            "There are 0 directories and 75 images in '10_food_classes_10_percent/train/grilled_salmon'.\n",
            "There are 0 directories and 75 images in '10_food_classes_10_percent/train/chicken_curry'.\n",
            "There are 0 directories and 75 images in '10_food_classes_10_percent/train/ramen'.\n",
            "There are 0 directories and 75 images in '10_food_classes_10_percent/train/pizza'.\n",
            "There are 0 directories and 75 images in '10_food_classes_10_percent/train/fried_rice'.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Create training and test directory paths\n",
        "train_dir = \"10_food_classes_10_percent/train\"\n",
        "test_dir = \"10_food_classes_10_percent/test\""
      ],
      "metadata": {
        "id": "rEHuVF9ivR1l"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import tensorflow as tf\n",
        "\n",
        "IMG_SIZE = (224, 224)\n",
        "BATCH_SIZE = 32\n",
        "train_data_10_percent = tf.keras.preprocessing.image_dataset_from_directory(directory=train_dir,\n",
        "                                                                            image_size=IMG_SIZE,\n",
        "                                                                            label_mode=\"categorical\",\n",
        "                                                                            batch_size=BATCH_SIZE)\n",
        "\n",
        "test_data = tf.keras.preprocessing.image_dataset_from_directory(directory=test_dir,\n",
        "                                                                image_size=IMG_SIZE,\n",
        "                                                                label_mode=\"categorical\",\n",
        "                                                                batch_size=BATCH_SIZE)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "DhJryQt9vlW-",
        "outputId": "813ab230-8717-4d37-dcbd-734c32d9d0c0"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Found 750 files belonging to 10 classes.\n",
            "Found 2500 files belonging to 10 classes.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "train_data_10_percent"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dG1mD-HowOcU",
        "outputId": "a88670f4-16cc-4476-8728-72ea31fcc740"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<BatchDataset element_spec=(TensorSpec(shape=(None, 224, 224, 3), dtype=tf.float32, name=None), TensorSpec(shape=(None, 10), dtype=tf.float32, name=None))>"
            ]
          },
          "metadata": {},
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Check out the class names of our dataset\n",
        "train_data_10_percent.class_names"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CEkYTHM_w1B5",
        "outputId": "6e873d1c-4857-43c8-dd95-8e6a1f26ec63"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['chicken_curry',\n",
              " 'chicken_wings',\n",
              " 'fried_rice',\n",
              " 'grilled_salmon',\n",
              " 'hamburger',\n",
              " 'ice_cream',\n",
              " 'pizza',\n",
              " 'ramen',\n",
              " 'steak',\n",
              " 'sushi']"
            ]
          },
          "metadata": {},
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# See an example of a batch of data\n",
        "for images, labels in train_data_10_percent.take(1):\n",
        "    print(images, labels)"
      ],
      "metadata": {
        "id": "QS62ib-_xXcS"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Model 0: Building a transfer learning feature extraction model using the Keras Functional API\n",
        "\n",
        "The sequential API is straight-forward, it runs our layers in sequential order.\n",
        "But the functional API gives us more flexibility with our models."
      ],
      "metadata": {
        "id": "Aw79HYGOxomA"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# 1. Create base model with tf.keras.applications\n",
        "base_model = tf.keras.applications.EfficientNetB0(include_top=False)\n",
        "\n",
        "# 2. Freeze the base model (so the underlying pre-trained patterns aren't updated during training)\n",
        "base_model.trainable = False\n",
        "\n",
        "# 3. Create inputs into our model\n",
        "inputs = tf.keras.layers.Input(shape=(224, 224, 3), name=\"input_layer\")\n",
        "\n",
        "# 4. If using ResNet50V2 you will need to normalize inputs (you don't have to for EfficientNet)\n",
        "# x = tf.keras.layers.experimental.preprocessing.Rescaling(1./255)(inputs)\n",
        "\n",
        "# 5. Pass the inputs to the base_model\n",
        "x = base_model(inputs)\n",
        "print(f\"Shape after passing inputs through base model: {x.shape}\")\n",
        "\n",
        "# 6. Average pool the outputs of the base model (aggregate all the most important information, reduce number of computations)\n",
        "x = tf.keras.layers.GlobalAveragePooling2D(name=\"global_average_pooling_layer\")(x)\n",
        "print(f\"Shape after GlobalAveragePooling2D: {x.shape}\")\n",
        "\n",
        "# 7. Create the output activation layer\n",
        "outputs = tf.keras.layers.Dense(10, activation=\"softmax\", name=\"output_layer\")(x)\n",
        "\n",
        "# 8. Combine the inputs with the outputs into a model\n",
        "model_0 = tf.keras.Model(inputs, outputs)\n",
        "\n",
        "# 9. Compile the model\n",
        "model_0.compile(loss=\"CategoricalCrossentropy\",\n",
        "                optimizer=\"Adam\",\n",
        "                metrics=[\"accuracy\"])\n",
        "\n",
        "# 10. Fit the model and save its history\n",
        "history_0_percent = model_0.fit(train_data_10_percent,\n",
        "                                epochs=5,\n",
        "                                steps_per_epoch=len(train_data_10_percent),\n",
        "                                validation_data=test_data,\n",
        "                                validation_steps=int(len(test_data)* 0.25),\n",
        "                                callbacks=[create_tensorboard_callback(\"transfer_learning\",\n",
        "                                                                       \"10_percent_feature_extraction\")])\n",
        "\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "aiVz3jqtiYJI",
        "outputId": "73b4d492-342b-48ad-ab21-e7b8cec79db0"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://storage.googleapis.com/keras-applications/efficientnetb0_notop.h5\n",
            "16705208/16705208 [==============================] - 0s 0us/step\n",
            "Shape after passing inputs through base model: (None, 7, 7, 1280)\n",
            "Shape after GlobalAveragePooling2D: (None, 1280)\n",
            "Saving TensorBoard log files to: transfer_learning/10_percent_feature_extraction/20230106-162448\n",
            "Epoch 1/5\n",
            "24/24 [==============================] - 19s 262ms/step - loss: 1.8501 - accuracy: 0.4320 - val_loss: 1.2863 - val_accuracy: 0.7155\n",
            "Epoch 2/5\n",
            "24/24 [==============================] - 5s 188ms/step - loss: 1.0748 - accuracy: 0.7720 - val_loss: 0.8506 - val_accuracy: 0.8257\n",
            "Epoch 3/5\n",
            "24/24 [==============================] - 4s 157ms/step - loss: 0.7870 - accuracy: 0.8173 - val_loss: 0.6991 - val_accuracy: 0.8536\n",
            "Epoch 4/5\n",
            "24/24 [==============================] - 4s 158ms/step - loss: 0.6277 - accuracy: 0.8613 - val_loss: 0.5950 - val_accuracy: 0.8569\n",
            "Epoch 5/5\n",
            "24/24 [==============================] - 5s 188ms/step - loss: 0.5346 - accuracy: 0.8840 - val_loss: 0.5525 - val_accuracy: 0.8717\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Evaluate on the full test dataset\n",
        "model_0.evaluate(test_data)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2hJ9NwwyldVY",
        "outputId": "17efc1e0-9666-4518-84b0-6926fae3f8ed"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "79/79 [==============================] - 7s 81ms/step - loss: 0.5510 - accuracy: 0.8612\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.5509842038154602, 0.8611999750137329]"
            ]
          },
          "metadata": {},
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Check the layers in our base model\n",
        "for layer_number, layer in enumerate(base_model.layers):\n",
        "  print(layer_number, layer.name)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JeILS-gtmhf2",
        "outputId": "f975348b-c07e-4442-eac9-7b557418eb6e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0 input_1\n",
            "1 rescaling\n",
            "2 normalization\n",
            "3 tf.math.truediv\n",
            "4 stem_conv_pad\n",
            "5 stem_conv\n",
            "6 stem_bn\n",
            "7 stem_activation\n",
            "8 block1a_dwconv\n",
            "9 block1a_bn\n",
            "10 block1a_activation\n",
            "11 block1a_se_squeeze\n",
            "12 block1a_se_reshape\n",
            "13 block1a_se_reduce\n",
            "14 block1a_se_expand\n",
            "15 block1a_se_excite\n",
            "16 block1a_project_conv\n",
            "17 block1a_project_bn\n",
            "18 block2a_expand_conv\n",
            "19 block2a_expand_bn\n",
            "20 block2a_expand_activation\n",
            "21 block2a_dwconv_pad\n",
            "22 block2a_dwconv\n",
            "23 block2a_bn\n",
            "24 block2a_activation\n",
            "25 block2a_se_squeeze\n",
            "26 block2a_se_reshape\n",
            "27 block2a_se_reduce\n",
            "28 block2a_se_expand\n",
            "29 block2a_se_excite\n",
            "30 block2a_project_conv\n",
            "31 block2a_project_bn\n",
            "32 block2b_expand_conv\n",
            "33 block2b_expand_bn\n",
            "34 block2b_expand_activation\n",
            "35 block2b_dwconv\n",
            "36 block2b_bn\n",
            "37 block2b_activation\n",
            "38 block2b_se_squeeze\n",
            "39 block2b_se_reshape\n",
            "40 block2b_se_reduce\n",
            "41 block2b_se_expand\n",
            "42 block2b_se_excite\n",
            "43 block2b_project_conv\n",
            "44 block2b_project_bn\n",
            "45 block2b_drop\n",
            "46 block2b_add\n",
            "47 block3a_expand_conv\n",
            "48 block3a_expand_bn\n",
            "49 block3a_expand_activation\n",
            "50 block3a_dwconv_pad\n",
            "51 block3a_dwconv\n",
            "52 block3a_bn\n",
            "53 block3a_activation\n",
            "54 block3a_se_squeeze\n",
            "55 block3a_se_reshape\n",
            "56 block3a_se_reduce\n",
            "57 block3a_se_expand\n",
            "58 block3a_se_excite\n",
            "59 block3a_project_conv\n",
            "60 block3a_project_bn\n",
            "61 block3b_expand_conv\n",
            "62 block3b_expand_bn\n",
            "63 block3b_expand_activation\n",
            "64 block3b_dwconv\n",
            "65 block3b_bn\n",
            "66 block3b_activation\n",
            "67 block3b_se_squeeze\n",
            "68 block3b_se_reshape\n",
            "69 block3b_se_reduce\n",
            "70 block3b_se_expand\n",
            "71 block3b_se_excite\n",
            "72 block3b_project_conv\n",
            "73 block3b_project_bn\n",
            "74 block3b_drop\n",
            "75 block3b_add\n",
            "76 block4a_expand_conv\n",
            "77 block4a_expand_bn\n",
            "78 block4a_expand_activation\n",
            "79 block4a_dwconv_pad\n",
            "80 block4a_dwconv\n",
            "81 block4a_bn\n",
            "82 block4a_activation\n",
            "83 block4a_se_squeeze\n",
            "84 block4a_se_reshape\n",
            "85 block4a_se_reduce\n",
            "86 block4a_se_expand\n",
            "87 block4a_se_excite\n",
            "88 block4a_project_conv\n",
            "89 block4a_project_bn\n",
            "90 block4b_expand_conv\n",
            "91 block4b_expand_bn\n",
            "92 block4b_expand_activation\n",
            "93 block4b_dwconv\n",
            "94 block4b_bn\n",
            "95 block4b_activation\n",
            "96 block4b_se_squeeze\n",
            "97 block4b_se_reshape\n",
            "98 block4b_se_reduce\n",
            "99 block4b_se_expand\n",
            "100 block4b_se_excite\n",
            "101 block4b_project_conv\n",
            "102 block4b_project_bn\n",
            "103 block4b_drop\n",
            "104 block4b_add\n",
            "105 block4c_expand_conv\n",
            "106 block4c_expand_bn\n",
            "107 block4c_expand_activation\n",
            "108 block4c_dwconv\n",
            "109 block4c_bn\n",
            "110 block4c_activation\n",
            "111 block4c_se_squeeze\n",
            "112 block4c_se_reshape\n",
            "113 block4c_se_reduce\n",
            "114 block4c_se_expand\n",
            "115 block4c_se_excite\n",
            "116 block4c_project_conv\n",
            "117 block4c_project_bn\n",
            "118 block4c_drop\n",
            "119 block4c_add\n",
            "120 block5a_expand_conv\n",
            "121 block5a_expand_bn\n",
            "122 block5a_expand_activation\n",
            "123 block5a_dwconv\n",
            "124 block5a_bn\n",
            "125 block5a_activation\n",
            "126 block5a_se_squeeze\n",
            "127 block5a_se_reshape\n",
            "128 block5a_se_reduce\n",
            "129 block5a_se_expand\n",
            "130 block5a_se_excite\n",
            "131 block5a_project_conv\n",
            "132 block5a_project_bn\n",
            "133 block5b_expand_conv\n",
            "134 block5b_expand_bn\n",
            "135 block5b_expand_activation\n",
            "136 block5b_dwconv\n",
            "137 block5b_bn\n",
            "138 block5b_activation\n",
            "139 block5b_se_squeeze\n",
            "140 block5b_se_reshape\n",
            "141 block5b_se_reduce\n",
            "142 block5b_se_expand\n",
            "143 block5b_se_excite\n",
            "144 block5b_project_conv\n",
            "145 block5b_project_bn\n",
            "146 block5b_drop\n",
            "147 block5b_add\n",
            "148 block5c_expand_conv\n",
            "149 block5c_expand_bn\n",
            "150 block5c_expand_activation\n",
            "151 block5c_dwconv\n",
            "152 block5c_bn\n",
            "153 block5c_activation\n",
            "154 block5c_se_squeeze\n",
            "155 block5c_se_reshape\n",
            "156 block5c_se_reduce\n",
            "157 block5c_se_expand\n",
            "158 block5c_se_excite\n",
            "159 block5c_project_conv\n",
            "160 block5c_project_bn\n",
            "161 block5c_drop\n",
            "162 block5c_add\n",
            "163 block6a_expand_conv\n",
            "164 block6a_expand_bn\n",
            "165 block6a_expand_activation\n",
            "166 block6a_dwconv_pad\n",
            "167 block6a_dwconv\n",
            "168 block6a_bn\n",
            "169 block6a_activation\n",
            "170 block6a_se_squeeze\n",
            "171 block6a_se_reshape\n",
            "172 block6a_se_reduce\n",
            "173 block6a_se_expand\n",
            "174 block6a_se_excite\n",
            "175 block6a_project_conv\n",
            "176 block6a_project_bn\n",
            "177 block6b_expand_conv\n",
            "178 block6b_expand_bn\n",
            "179 block6b_expand_activation\n",
            "180 block6b_dwconv\n",
            "181 block6b_bn\n",
            "182 block6b_activation\n",
            "183 block6b_se_squeeze\n",
            "184 block6b_se_reshape\n",
            "185 block6b_se_reduce\n",
            "186 block6b_se_expand\n",
            "187 block6b_se_excite\n",
            "188 block6b_project_conv\n",
            "189 block6b_project_bn\n",
            "190 block6b_drop\n",
            "191 block6b_add\n",
            "192 block6c_expand_conv\n",
            "193 block6c_expand_bn\n",
            "194 block6c_expand_activation\n",
            "195 block6c_dwconv\n",
            "196 block6c_bn\n",
            "197 block6c_activation\n",
            "198 block6c_se_squeeze\n",
            "199 block6c_se_reshape\n",
            "200 block6c_se_reduce\n",
            "201 block6c_se_expand\n",
            "202 block6c_se_excite\n",
            "203 block6c_project_conv\n",
            "204 block6c_project_bn\n",
            "205 block6c_drop\n",
            "206 block6c_add\n",
            "207 block6d_expand_conv\n",
            "208 block6d_expand_bn\n",
            "209 block6d_expand_activation\n",
            "210 block6d_dwconv\n",
            "211 block6d_bn\n",
            "212 block6d_activation\n",
            "213 block6d_se_squeeze\n",
            "214 block6d_se_reshape\n",
            "215 block6d_se_reduce\n",
            "216 block6d_se_expand\n",
            "217 block6d_se_excite\n",
            "218 block6d_project_conv\n",
            "219 block6d_project_bn\n",
            "220 block6d_drop\n",
            "221 block6d_add\n",
            "222 block7a_expand_conv\n",
            "223 block7a_expand_bn\n",
            "224 block7a_expand_activation\n",
            "225 block7a_dwconv\n",
            "226 block7a_bn\n",
            "227 block7a_activation\n",
            "228 block7a_se_squeeze\n",
            "229 block7a_se_reshape\n",
            "230 block7a_se_reduce\n",
            "231 block7a_se_expand\n",
            "232 block7a_se_excite\n",
            "233 block7a_project_conv\n",
            "234 block7a_project_bn\n",
            "235 top_conv\n",
            "236 top_bn\n",
            "237 top_activation\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# How about we get a summary of the base model?\n",
        "base_model.summary()"
      ],
      "metadata": {
        "id": "mrf7d6UYmzUv",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "c311c5db-c3ff-4519-e6f4-9fc41e92e35e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"efficientnetb0\"\n",
            "__________________________________________________________________________________________________\n",
            " Layer (type)                   Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            " input_1 (InputLayer)           [(None, None, None,  0           []                               \n",
            "                                 3)]                                                              \n",
            "                                                                                                  \n",
            " rescaling (Rescaling)          (None, None, None,   0           ['input_1[0][0]']                \n",
            "                                3)                                                                \n",
            "                                                                                                  \n",
            " normalization (Normalization)  (None, None, None,   7           ['rescaling[0][0]']              \n",
            "                                3)                                                                \n",
            "                                                                                                  \n",
            " tf.math.truediv (TFOpLambda)   (None, None, None,   0           ['normalization[0][0]']          \n",
            "                                3)                                                                \n",
            "                                                                                                  \n",
            " stem_conv_pad (ZeroPadding2D)  (None, None, None,   0           ['tf.math.truediv[0][0]']        \n",
            "                                3)                                                                \n",
            "                                                                                                  \n",
            " stem_conv (Conv2D)             (None, None, None,   864         ['stem_conv_pad[0][0]']          \n",
            "                                32)                                                               \n",
            "                                                                                                  \n",
            " stem_bn (BatchNormalization)   (None, None, None,   128         ['stem_conv[0][0]']              \n",
            "                                32)                                                               \n",
            "                                                                                                  \n",
            " stem_activation (Activation)   (None, None, None,   0           ['stem_bn[0][0]']                \n",
            "                                32)                                                               \n",
            "                                                                                                  \n",
            " block1a_dwconv (DepthwiseConv2  (None, None, None,   288        ['stem_activation[0][0]']        \n",
            " D)                             32)                                                               \n",
            "                                                                                                  \n",
            " block1a_bn (BatchNormalization  (None, None, None,   128        ['block1a_dwconv[0][0]']         \n",
            " )                              32)                                                               \n",
            "                                                                                                  \n",
            " block1a_activation (Activation  (None, None, None,   0          ['block1a_bn[0][0]']             \n",
            " )                              32)                                                               \n",
            "                                                                                                  \n",
            " block1a_se_squeeze (GlobalAver  (None, 32)          0           ['block1a_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block1a_se_reshape (Reshape)   (None, 1, 1, 32)     0           ['block1a_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block1a_se_reduce (Conv2D)     (None, 1, 1, 8)      264         ['block1a_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block1a_se_expand (Conv2D)     (None, 1, 1, 32)     288         ['block1a_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block1a_se_excite (Multiply)   (None, None, None,   0           ['block1a_activation[0][0]',     \n",
            "                                32)                               'block1a_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block1a_project_conv (Conv2D)  (None, None, None,   512         ['block1a_se_excite[0][0]']      \n",
            "                                16)                                                               \n",
            "                                                                                                  \n",
            " block1a_project_bn (BatchNorma  (None, None, None,   64         ['block1a_project_conv[0][0]']   \n",
            " lization)                      16)                                                               \n",
            "                                                                                                  \n",
            " block2a_expand_conv (Conv2D)   (None, None, None,   1536        ['block1a_project_bn[0][0]']     \n",
            "                                96)                                                               \n",
            "                                                                                                  \n",
            " block2a_expand_bn (BatchNormal  (None, None, None,   384        ['block2a_expand_conv[0][0]']    \n",
            " ization)                       96)                                                               \n",
            "                                                                                                  \n",
            " block2a_expand_activation (Act  (None, None, None,   0          ['block2a_expand_bn[0][0]']      \n",
            " ivation)                       96)                                                               \n",
            "                                                                                                  \n",
            " block2a_dwconv_pad (ZeroPaddin  (None, None, None,   0          ['block2a_expand_activation[0][0]\n",
            " g2D)                           96)                              ']                               \n",
            "                                                                                                  \n",
            " block2a_dwconv (DepthwiseConv2  (None, None, None,   864        ['block2a_dwconv_pad[0][0]']     \n",
            " D)                             96)                                                               \n",
            "                                                                                                  \n",
            " block2a_bn (BatchNormalization  (None, None, None,   384        ['block2a_dwconv[0][0]']         \n",
            " )                              96)                                                               \n",
            "                                                                                                  \n",
            " block2a_activation (Activation  (None, None, None,   0          ['block2a_bn[0][0]']             \n",
            " )                              96)                                                               \n",
            "                                                                                                  \n",
            " block2a_se_squeeze (GlobalAver  (None, 96)          0           ['block2a_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block2a_se_reshape (Reshape)   (None, 1, 1, 96)     0           ['block2a_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block2a_se_reduce (Conv2D)     (None, 1, 1, 4)      388         ['block2a_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block2a_se_expand (Conv2D)     (None, 1, 1, 96)     480         ['block2a_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block2a_se_excite (Multiply)   (None, None, None,   0           ['block2a_activation[0][0]',     \n",
            "                                96)                               'block2a_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block2a_project_conv (Conv2D)  (None, None, None,   2304        ['block2a_se_excite[0][0]']      \n",
            "                                24)                                                               \n",
            "                                                                                                  \n",
            " block2a_project_bn (BatchNorma  (None, None, None,   96         ['block2a_project_conv[0][0]']   \n",
            " lization)                      24)                                                               \n",
            "                                                                                                  \n",
            " block2b_expand_conv (Conv2D)   (None, None, None,   3456        ['block2a_project_bn[0][0]']     \n",
            "                                144)                                                              \n",
            "                                                                                                  \n",
            " block2b_expand_bn (BatchNormal  (None, None, None,   576        ['block2b_expand_conv[0][0]']    \n",
            " ization)                       144)                                                              \n",
            "                                                                                                  \n",
            " block2b_expand_activation (Act  (None, None, None,   0          ['block2b_expand_bn[0][0]']      \n",
            " ivation)                       144)                                                              \n",
            "                                                                                                  \n",
            " block2b_dwconv (DepthwiseConv2  (None, None, None,   1296       ['block2b_expand_activation[0][0]\n",
            " D)                             144)                             ']                               \n",
            "                                                                                                  \n",
            " block2b_bn (BatchNormalization  (None, None, None,   576        ['block2b_dwconv[0][0]']         \n",
            " )                              144)                                                              \n",
            "                                                                                                  \n",
            " block2b_activation (Activation  (None, None, None,   0          ['block2b_bn[0][0]']             \n",
            " )                              144)                                                              \n",
            "                                                                                                  \n",
            " block2b_se_squeeze (GlobalAver  (None, 144)         0           ['block2b_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block2b_se_reshape (Reshape)   (None, 1, 1, 144)    0           ['block2b_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block2b_se_reduce (Conv2D)     (None, 1, 1, 6)      870         ['block2b_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block2b_se_expand (Conv2D)     (None, 1, 1, 144)    1008        ['block2b_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block2b_se_excite (Multiply)   (None, None, None,   0           ['block2b_activation[0][0]',     \n",
            "                                144)                              'block2b_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block2b_project_conv (Conv2D)  (None, None, None,   3456        ['block2b_se_excite[0][0]']      \n",
            "                                24)                                                               \n",
            "                                                                                                  \n",
            " block2b_project_bn (BatchNorma  (None, None, None,   96         ['block2b_project_conv[0][0]']   \n",
            " lization)                      24)                                                               \n",
            "                                                                                                  \n",
            " block2b_drop (Dropout)         (None, None, None,   0           ['block2b_project_bn[0][0]']     \n",
            "                                24)                                                               \n",
            "                                                                                                  \n",
            " block2b_add (Add)              (None, None, None,   0           ['block2b_drop[0][0]',           \n",
            "                                24)                               'block2a_project_bn[0][0]']     \n",
            "                                                                                                  \n",
            " block3a_expand_conv (Conv2D)   (None, None, None,   3456        ['block2b_add[0][0]']            \n",
            "                                144)                                                              \n",
            "                                                                                                  \n",
            " block3a_expand_bn (BatchNormal  (None, None, None,   576        ['block3a_expand_conv[0][0]']    \n",
            " ization)                       144)                                                              \n",
            "                                                                                                  \n",
            " block3a_expand_activation (Act  (None, None, None,   0          ['block3a_expand_bn[0][0]']      \n",
            " ivation)                       144)                                                              \n",
            "                                                                                                  \n",
            " block3a_dwconv_pad (ZeroPaddin  (None, None, None,   0          ['block3a_expand_activation[0][0]\n",
            " g2D)                           144)                             ']                               \n",
            "                                                                                                  \n",
            " block3a_dwconv (DepthwiseConv2  (None, None, None,   3600       ['block3a_dwconv_pad[0][0]']     \n",
            " D)                             144)                                                              \n",
            "                                                                                                  \n",
            " block3a_bn (BatchNormalization  (None, None, None,   576        ['block3a_dwconv[0][0]']         \n",
            " )                              144)                                                              \n",
            "                                                                                                  \n",
            " block3a_activation (Activation  (None, None, None,   0          ['block3a_bn[0][0]']             \n",
            " )                              144)                                                              \n",
            "                                                                                                  \n",
            " block3a_se_squeeze (GlobalAver  (None, 144)         0           ['block3a_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block3a_se_reshape (Reshape)   (None, 1, 1, 144)    0           ['block3a_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block3a_se_reduce (Conv2D)     (None, 1, 1, 6)      870         ['block3a_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block3a_se_expand (Conv2D)     (None, 1, 1, 144)    1008        ['block3a_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block3a_se_excite (Multiply)   (None, None, None,   0           ['block3a_activation[0][0]',     \n",
            "                                144)                              'block3a_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block3a_project_conv (Conv2D)  (None, None, None,   5760        ['block3a_se_excite[0][0]']      \n",
            "                                40)                                                               \n",
            "                                                                                                  \n",
            " block3a_project_bn (BatchNorma  (None, None, None,   160        ['block3a_project_conv[0][0]']   \n",
            " lization)                      40)                                                               \n",
            "                                                                                                  \n",
            " block3b_expand_conv (Conv2D)   (None, None, None,   9600        ['block3a_project_bn[0][0]']     \n",
            "                                240)                                                              \n",
            "                                                                                                  \n",
            " block3b_expand_bn (BatchNormal  (None, None, None,   960        ['block3b_expand_conv[0][0]']    \n",
            " ization)                       240)                                                              \n",
            "                                                                                                  \n",
            " block3b_expand_activation (Act  (None, None, None,   0          ['block3b_expand_bn[0][0]']      \n",
            " ivation)                       240)                                                              \n",
            "                                                                                                  \n",
            " block3b_dwconv (DepthwiseConv2  (None, None, None,   6000       ['block3b_expand_activation[0][0]\n",
            " D)                             240)                             ']                               \n",
            "                                                                                                  \n",
            " block3b_bn (BatchNormalization  (None, None, None,   960        ['block3b_dwconv[0][0]']         \n",
            " )                              240)                                                              \n",
            "                                                                                                  \n",
            " block3b_activation (Activation  (None, None, None,   0          ['block3b_bn[0][0]']             \n",
            " )                              240)                                                              \n",
            "                                                                                                  \n",
            " block3b_se_squeeze (GlobalAver  (None, 240)         0           ['block3b_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block3b_se_reshape (Reshape)   (None, 1, 1, 240)    0           ['block3b_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block3b_se_reduce (Conv2D)     (None, 1, 1, 10)     2410        ['block3b_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block3b_se_expand (Conv2D)     (None, 1, 1, 240)    2640        ['block3b_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block3b_se_excite (Multiply)   (None, None, None,   0           ['block3b_activation[0][0]',     \n",
            "                                240)                              'block3b_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block3b_project_conv (Conv2D)  (None, None, None,   9600        ['block3b_se_excite[0][0]']      \n",
            "                                40)                                                               \n",
            "                                                                                                  \n",
            " block3b_project_bn (BatchNorma  (None, None, None,   160        ['block3b_project_conv[0][0]']   \n",
            " lization)                      40)                                                               \n",
            "                                                                                                  \n",
            " block3b_drop (Dropout)         (None, None, None,   0           ['block3b_project_bn[0][0]']     \n",
            "                                40)                                                               \n",
            "                                                                                                  \n",
            " block3b_add (Add)              (None, None, None,   0           ['block3b_drop[0][0]',           \n",
            "                                40)                               'block3a_project_bn[0][0]']     \n",
            "                                                                                                  \n",
            " block4a_expand_conv (Conv2D)   (None, None, None,   9600        ['block3b_add[0][0]']            \n",
            "                                240)                                                              \n",
            "                                                                                                  \n",
            " block4a_expand_bn (BatchNormal  (None, None, None,   960        ['block4a_expand_conv[0][0]']    \n",
            " ization)                       240)                                                              \n",
            "                                                                                                  \n",
            " block4a_expand_activation (Act  (None, None, None,   0          ['block4a_expand_bn[0][0]']      \n",
            " ivation)                       240)                                                              \n",
            "                                                                                                  \n",
            " block4a_dwconv_pad (ZeroPaddin  (None, None, None,   0          ['block4a_expand_activation[0][0]\n",
            " g2D)                           240)                             ']                               \n",
            "                                                                                                  \n",
            " block4a_dwconv (DepthwiseConv2  (None, None, None,   2160       ['block4a_dwconv_pad[0][0]']     \n",
            " D)                             240)                                                              \n",
            "                                                                                                  \n",
            " block4a_bn (BatchNormalization  (None, None, None,   960        ['block4a_dwconv[0][0]']         \n",
            " )                              240)                                                              \n",
            "                                                                                                  \n",
            " block4a_activation (Activation  (None, None, None,   0          ['block4a_bn[0][0]']             \n",
            " )                              240)                                                              \n",
            "                                                                                                  \n",
            " block4a_se_squeeze (GlobalAver  (None, 240)         0           ['block4a_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block4a_se_reshape (Reshape)   (None, 1, 1, 240)    0           ['block4a_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block4a_se_reduce (Conv2D)     (None, 1, 1, 10)     2410        ['block4a_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block4a_se_expand (Conv2D)     (None, 1, 1, 240)    2640        ['block4a_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block4a_se_excite (Multiply)   (None, None, None,   0           ['block4a_activation[0][0]',     \n",
            "                                240)                              'block4a_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block4a_project_conv (Conv2D)  (None, None, None,   19200       ['block4a_se_excite[0][0]']      \n",
            "                                80)                                                               \n",
            "                                                                                                  \n",
            " block4a_project_bn (BatchNorma  (None, None, None,   320        ['block4a_project_conv[0][0]']   \n",
            " lization)                      80)                                                               \n",
            "                                                                                                  \n",
            " block4b_expand_conv (Conv2D)   (None, None, None,   38400       ['block4a_project_bn[0][0]']     \n",
            "                                480)                                                              \n",
            "                                                                                                  \n",
            " block4b_expand_bn (BatchNormal  (None, None, None,   1920       ['block4b_expand_conv[0][0]']    \n",
            " ization)                       480)                                                              \n",
            "                                                                                                  \n",
            " block4b_expand_activation (Act  (None, None, None,   0          ['block4b_expand_bn[0][0]']      \n",
            " ivation)                       480)                                                              \n",
            "                                                                                                  \n",
            " block4b_dwconv (DepthwiseConv2  (None, None, None,   4320       ['block4b_expand_activation[0][0]\n",
            " D)                             480)                             ']                               \n",
            "                                                                                                  \n",
            " block4b_bn (BatchNormalization  (None, None, None,   1920       ['block4b_dwconv[0][0]']         \n",
            " )                              480)                                                              \n",
            "                                                                                                  \n",
            " block4b_activation (Activation  (None, None, None,   0          ['block4b_bn[0][0]']             \n",
            " )                              480)                                                              \n",
            "                                                                                                  \n",
            " block4b_se_squeeze (GlobalAver  (None, 480)         0           ['block4b_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block4b_se_reshape (Reshape)   (None, 1, 1, 480)    0           ['block4b_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block4b_se_reduce (Conv2D)     (None, 1, 1, 20)     9620        ['block4b_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block4b_se_expand (Conv2D)     (None, 1, 1, 480)    10080       ['block4b_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block4b_se_excite (Multiply)   (None, None, None,   0           ['block4b_activation[0][0]',     \n",
            "                                480)                              'block4b_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block4b_project_conv (Conv2D)  (None, None, None,   38400       ['block4b_se_excite[0][0]']      \n",
            "                                80)                                                               \n",
            "                                                                                                  \n",
            " block4b_project_bn (BatchNorma  (None, None, None,   320        ['block4b_project_conv[0][0]']   \n",
            " lization)                      80)                                                               \n",
            "                                                                                                  \n",
            " block4b_drop (Dropout)         (None, None, None,   0           ['block4b_project_bn[0][0]']     \n",
            "                                80)                                                               \n",
            "                                                                                                  \n",
            " block4b_add (Add)              (None, None, None,   0           ['block4b_drop[0][0]',           \n",
            "                                80)                               'block4a_project_bn[0][0]']     \n",
            "                                                                                                  \n",
            " block4c_expand_conv (Conv2D)   (None, None, None,   38400       ['block4b_add[0][0]']            \n",
            "                                480)                                                              \n",
            "                                                                                                  \n",
            " block4c_expand_bn (BatchNormal  (None, None, None,   1920       ['block4c_expand_conv[0][0]']    \n",
            " ization)                       480)                                                              \n",
            "                                                                                                  \n",
            " block4c_expand_activation (Act  (None, None, None,   0          ['block4c_expand_bn[0][0]']      \n",
            " ivation)                       480)                                                              \n",
            "                                                                                                  \n",
            " block4c_dwconv (DepthwiseConv2  (None, None, None,   4320       ['block4c_expand_activation[0][0]\n",
            " D)                             480)                             ']                               \n",
            "                                                                                                  \n",
            " block4c_bn (BatchNormalization  (None, None, None,   1920       ['block4c_dwconv[0][0]']         \n",
            " )                              480)                                                              \n",
            "                                                                                                  \n",
            " block4c_activation (Activation  (None, None, None,   0          ['block4c_bn[0][0]']             \n",
            " )                              480)                                                              \n",
            "                                                                                                  \n",
            " block4c_se_squeeze (GlobalAver  (None, 480)         0           ['block4c_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block4c_se_reshape (Reshape)   (None, 1, 1, 480)    0           ['block4c_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block4c_se_reduce (Conv2D)     (None, 1, 1, 20)     9620        ['block4c_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block4c_se_expand (Conv2D)     (None, 1, 1, 480)    10080       ['block4c_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block4c_se_excite (Multiply)   (None, None, None,   0           ['block4c_activation[0][0]',     \n",
            "                                480)                              'block4c_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block4c_project_conv (Conv2D)  (None, None, None,   38400       ['block4c_se_excite[0][0]']      \n",
            "                                80)                                                               \n",
            "                                                                                                  \n",
            " block4c_project_bn (BatchNorma  (None, None, None,   320        ['block4c_project_conv[0][0]']   \n",
            " lization)                      80)                                                               \n",
            "                                                                                                  \n",
            " block4c_drop (Dropout)         (None, None, None,   0           ['block4c_project_bn[0][0]']     \n",
            "                                80)                                                               \n",
            "                                                                                                  \n",
            " block4c_add (Add)              (None, None, None,   0           ['block4c_drop[0][0]',           \n",
            "                                80)                               'block4b_add[0][0]']            \n",
            "                                                                                                  \n",
            " block5a_expand_conv (Conv2D)   (None, None, None,   38400       ['block4c_add[0][0]']            \n",
            "                                480)                                                              \n",
            "                                                                                                  \n",
            " block5a_expand_bn (BatchNormal  (None, None, None,   1920       ['block5a_expand_conv[0][0]']    \n",
            " ization)                       480)                                                              \n",
            "                                                                                                  \n",
            " block5a_expand_activation (Act  (None, None, None,   0          ['block5a_expand_bn[0][0]']      \n",
            " ivation)                       480)                                                              \n",
            "                                                                                                  \n",
            " block5a_dwconv (DepthwiseConv2  (None, None, None,   12000      ['block5a_expand_activation[0][0]\n",
            " D)                             480)                             ']                               \n",
            "                                                                                                  \n",
            " block5a_bn (BatchNormalization  (None, None, None,   1920       ['block5a_dwconv[0][0]']         \n",
            " )                              480)                                                              \n",
            "                                                                                                  \n",
            " block5a_activation (Activation  (None, None, None,   0          ['block5a_bn[0][0]']             \n",
            " )                              480)                                                              \n",
            "                                                                                                  \n",
            " block5a_se_squeeze (GlobalAver  (None, 480)         0           ['block5a_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block5a_se_reshape (Reshape)   (None, 1, 1, 480)    0           ['block5a_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block5a_se_reduce (Conv2D)     (None, 1, 1, 20)     9620        ['block5a_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block5a_se_expand (Conv2D)     (None, 1, 1, 480)    10080       ['block5a_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block5a_se_excite (Multiply)   (None, None, None,   0           ['block5a_activation[0][0]',     \n",
            "                                480)                              'block5a_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block5a_project_conv (Conv2D)  (None, None, None,   53760       ['block5a_se_excite[0][0]']      \n",
            "                                112)                                                              \n",
            "                                                                                                  \n",
            " block5a_project_bn (BatchNorma  (None, None, None,   448        ['block5a_project_conv[0][0]']   \n",
            " lization)                      112)                                                              \n",
            "                                                                                                  \n",
            " block5b_expand_conv (Conv2D)   (None, None, None,   75264       ['block5a_project_bn[0][0]']     \n",
            "                                672)                                                              \n",
            "                                                                                                  \n",
            " block5b_expand_bn (BatchNormal  (None, None, None,   2688       ['block5b_expand_conv[0][0]']    \n",
            " ization)                       672)                                                              \n",
            "                                                                                                  \n",
            " block5b_expand_activation (Act  (None, None, None,   0          ['block5b_expand_bn[0][0]']      \n",
            " ivation)                       672)                                                              \n",
            "                                                                                                  \n",
            " block5b_dwconv (DepthwiseConv2  (None, None, None,   16800      ['block5b_expand_activation[0][0]\n",
            " D)                             672)                             ']                               \n",
            "                                                                                                  \n",
            " block5b_bn (BatchNormalization  (None, None, None,   2688       ['block5b_dwconv[0][0]']         \n",
            " )                              672)                                                              \n",
            "                                                                                                  \n",
            " block5b_activation (Activation  (None, None, None,   0          ['block5b_bn[0][0]']             \n",
            " )                              672)                                                              \n",
            "                                                                                                  \n",
            " block5b_se_squeeze (GlobalAver  (None, 672)         0           ['block5b_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block5b_se_reshape (Reshape)   (None, 1, 1, 672)    0           ['block5b_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block5b_se_reduce (Conv2D)     (None, 1, 1, 28)     18844       ['block5b_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block5b_se_expand (Conv2D)     (None, 1, 1, 672)    19488       ['block5b_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block5b_se_excite (Multiply)   (None, None, None,   0           ['block5b_activation[0][0]',     \n",
            "                                672)                              'block5b_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block5b_project_conv (Conv2D)  (None, None, None,   75264       ['block5b_se_excite[0][0]']      \n",
            "                                112)                                                              \n",
            "                                                                                                  \n",
            " block5b_project_bn (BatchNorma  (None, None, None,   448        ['block5b_project_conv[0][0]']   \n",
            " lization)                      112)                                                              \n",
            "                                                                                                  \n",
            " block5b_drop (Dropout)         (None, None, None,   0           ['block5b_project_bn[0][0]']     \n",
            "                                112)                                                              \n",
            "                                                                                                  \n",
            " block5b_add (Add)              (None, None, None,   0           ['block5b_drop[0][0]',           \n",
            "                                112)                              'block5a_project_bn[0][0]']     \n",
            "                                                                                                  \n",
            " block5c_expand_conv (Conv2D)   (None, None, None,   75264       ['block5b_add[0][0]']            \n",
            "                                672)                                                              \n",
            "                                                                                                  \n",
            " block5c_expand_bn (BatchNormal  (None, None, None,   2688       ['block5c_expand_conv[0][0]']    \n",
            " ization)                       672)                                                              \n",
            "                                                                                                  \n",
            " block5c_expand_activation (Act  (None, None, None,   0          ['block5c_expand_bn[0][0]']      \n",
            " ivation)                       672)                                                              \n",
            "                                                                                                  \n",
            " block5c_dwconv (DepthwiseConv2  (None, None, None,   16800      ['block5c_expand_activation[0][0]\n",
            " D)                             672)                             ']                               \n",
            "                                                                                                  \n",
            " block5c_bn (BatchNormalization  (None, None, None,   2688       ['block5c_dwconv[0][0]']         \n",
            " )                              672)                                                              \n",
            "                                                                                                  \n",
            " block5c_activation (Activation  (None, None, None,   0          ['block5c_bn[0][0]']             \n",
            " )                              672)                                                              \n",
            "                                                                                                  \n",
            " block5c_se_squeeze (GlobalAver  (None, 672)         0           ['block5c_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block5c_se_reshape (Reshape)   (None, 1, 1, 672)    0           ['block5c_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block5c_se_reduce (Conv2D)     (None, 1, 1, 28)     18844       ['block5c_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block5c_se_expand (Conv2D)     (None, 1, 1, 672)    19488       ['block5c_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block5c_se_excite (Multiply)   (None, None, None,   0           ['block5c_activation[0][0]',     \n",
            "                                672)                              'block5c_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block5c_project_conv (Conv2D)  (None, None, None,   75264       ['block5c_se_excite[0][0]']      \n",
            "                                112)                                                              \n",
            "                                                                                                  \n",
            " block5c_project_bn (BatchNorma  (None, None, None,   448        ['block5c_project_conv[0][0]']   \n",
            " lization)                      112)                                                              \n",
            "                                                                                                  \n",
            " block5c_drop (Dropout)         (None, None, None,   0           ['block5c_project_bn[0][0]']     \n",
            "                                112)                                                              \n",
            "                                                                                                  \n",
            " block5c_add (Add)              (None, None, None,   0           ['block5c_drop[0][0]',           \n",
            "                                112)                              'block5b_add[0][0]']            \n",
            "                                                                                                  \n",
            " block6a_expand_conv (Conv2D)   (None, None, None,   75264       ['block5c_add[0][0]']            \n",
            "                                672)                                                              \n",
            "                                                                                                  \n",
            " block6a_expand_bn (BatchNormal  (None, None, None,   2688       ['block6a_expand_conv[0][0]']    \n",
            " ization)                       672)                                                              \n",
            "                                                                                                  \n",
            " block6a_expand_activation (Act  (None, None, None,   0          ['block6a_expand_bn[0][0]']      \n",
            " ivation)                       672)                                                              \n",
            "                                                                                                  \n",
            " block6a_dwconv_pad (ZeroPaddin  (None, None, None,   0          ['block6a_expand_activation[0][0]\n",
            " g2D)                           672)                             ']                               \n",
            "                                                                                                  \n",
            " block6a_dwconv (DepthwiseConv2  (None, None, None,   16800      ['block6a_dwconv_pad[0][0]']     \n",
            " D)                             672)                                                              \n",
            "                                                                                                  \n",
            " block6a_bn (BatchNormalization  (None, None, None,   2688       ['block6a_dwconv[0][0]']         \n",
            " )                              672)                                                              \n",
            "                                                                                                  \n",
            " block6a_activation (Activation  (None, None, None,   0          ['block6a_bn[0][0]']             \n",
            " )                              672)                                                              \n",
            "                                                                                                  \n",
            " block6a_se_squeeze (GlobalAver  (None, 672)         0           ['block6a_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block6a_se_reshape (Reshape)   (None, 1, 1, 672)    0           ['block6a_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block6a_se_reduce (Conv2D)     (None, 1, 1, 28)     18844       ['block6a_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block6a_se_expand (Conv2D)     (None, 1, 1, 672)    19488       ['block6a_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block6a_se_excite (Multiply)   (None, None, None,   0           ['block6a_activation[0][0]',     \n",
            "                                672)                              'block6a_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block6a_project_conv (Conv2D)  (None, None, None,   129024      ['block6a_se_excite[0][0]']      \n",
            "                                192)                                                              \n",
            "                                                                                                  \n",
            " block6a_project_bn (BatchNorma  (None, None, None,   768        ['block6a_project_conv[0][0]']   \n",
            " lization)                      192)                                                              \n",
            "                                                                                                  \n",
            " block6b_expand_conv (Conv2D)   (None, None, None,   221184      ['block6a_project_bn[0][0]']     \n",
            "                                1152)                                                             \n",
            "                                                                                                  \n",
            " block6b_expand_bn (BatchNormal  (None, None, None,   4608       ['block6b_expand_conv[0][0]']    \n",
            " ization)                       1152)                                                             \n",
            "                                                                                                  \n",
            " block6b_expand_activation (Act  (None, None, None,   0          ['block6b_expand_bn[0][0]']      \n",
            " ivation)                       1152)                                                             \n",
            "                                                                                                  \n",
            " block6b_dwconv (DepthwiseConv2  (None, None, None,   28800      ['block6b_expand_activation[0][0]\n",
            " D)                             1152)                            ']                               \n",
            "                                                                                                  \n",
            " block6b_bn (BatchNormalization  (None, None, None,   4608       ['block6b_dwconv[0][0]']         \n",
            " )                              1152)                                                             \n",
            "                                                                                                  \n",
            " block6b_activation (Activation  (None, None, None,   0          ['block6b_bn[0][0]']             \n",
            " )                              1152)                                                             \n",
            "                                                                                                  \n",
            " block6b_se_squeeze (GlobalAver  (None, 1152)        0           ['block6b_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block6b_se_reshape (Reshape)   (None, 1, 1, 1152)   0           ['block6b_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block6b_se_reduce (Conv2D)     (None, 1, 1, 48)     55344       ['block6b_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block6b_se_expand (Conv2D)     (None, 1, 1, 1152)   56448       ['block6b_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block6b_se_excite (Multiply)   (None, None, None,   0           ['block6b_activation[0][0]',     \n",
            "                                1152)                             'block6b_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block6b_project_conv (Conv2D)  (None, None, None,   221184      ['block6b_se_excite[0][0]']      \n",
            "                                192)                                                              \n",
            "                                                                                                  \n",
            " block6b_project_bn (BatchNorma  (None, None, None,   768        ['block6b_project_conv[0][0]']   \n",
            " lization)                      192)                                                              \n",
            "                                                                                                  \n",
            " block6b_drop (Dropout)         (None, None, None,   0           ['block6b_project_bn[0][0]']     \n",
            "                                192)                                                              \n",
            "                                                                                                  \n",
            " block6b_add (Add)              (None, None, None,   0           ['block6b_drop[0][0]',           \n",
            "                                192)                              'block6a_project_bn[0][0]']     \n",
            "                                                                                                  \n",
            " block6c_expand_conv (Conv2D)   (None, None, None,   221184      ['block6b_add[0][0]']            \n",
            "                                1152)                                                             \n",
            "                                                                                                  \n",
            " block6c_expand_bn (BatchNormal  (None, None, None,   4608       ['block6c_expand_conv[0][0]']    \n",
            " ization)                       1152)                                                             \n",
            "                                                                                                  \n",
            " block6c_expand_activation (Act  (None, None, None,   0          ['block6c_expand_bn[0][0]']      \n",
            " ivation)                       1152)                                                             \n",
            "                                                                                                  \n",
            " block6c_dwconv (DepthwiseConv2  (None, None, None,   28800      ['block6c_expand_activation[0][0]\n",
            " D)                             1152)                            ']                               \n",
            "                                                                                                  \n",
            " block6c_bn (BatchNormalization  (None, None, None,   4608       ['block6c_dwconv[0][0]']         \n",
            " )                              1152)                                                             \n",
            "                                                                                                  \n",
            " block6c_activation (Activation  (None, None, None,   0          ['block6c_bn[0][0]']             \n",
            " )                              1152)                                                             \n",
            "                                                                                                  \n",
            " block6c_se_squeeze (GlobalAver  (None, 1152)        0           ['block6c_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block6c_se_reshape (Reshape)   (None, 1, 1, 1152)   0           ['block6c_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block6c_se_reduce (Conv2D)     (None, 1, 1, 48)     55344       ['block6c_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block6c_se_expand (Conv2D)     (None, 1, 1, 1152)   56448       ['block6c_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block6c_se_excite (Multiply)   (None, None, None,   0           ['block6c_activation[0][0]',     \n",
            "                                1152)                             'block6c_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block6c_project_conv (Conv2D)  (None, None, None,   221184      ['block6c_se_excite[0][0]']      \n",
            "                                192)                                                              \n",
            "                                                                                                  \n",
            " block6c_project_bn (BatchNorma  (None, None, None,   768        ['block6c_project_conv[0][0]']   \n",
            " lization)                      192)                                                              \n",
            "                                                                                                  \n",
            " block6c_drop (Dropout)         (None, None, None,   0           ['block6c_project_bn[0][0]']     \n",
            "                                192)                                                              \n",
            "                                                                                                  \n",
            " block6c_add (Add)              (None, None, None,   0           ['block6c_drop[0][0]',           \n",
            "                                192)                              'block6b_add[0][0]']            \n",
            "                                                                                                  \n",
            " block6d_expand_conv (Conv2D)   (None, None, None,   221184      ['block6c_add[0][0]']            \n",
            "                                1152)                                                             \n",
            "                                                                                                  \n",
            " block6d_expand_bn (BatchNormal  (None, None, None,   4608       ['block6d_expand_conv[0][0]']    \n",
            " ization)                       1152)                                                             \n",
            "                                                                                                  \n",
            " block6d_expand_activation (Act  (None, None, None,   0          ['block6d_expand_bn[0][0]']      \n",
            " ivation)                       1152)                                                             \n",
            "                                                                                                  \n",
            " block6d_dwconv (DepthwiseConv2  (None, None, None,   28800      ['block6d_expand_activation[0][0]\n",
            " D)                             1152)                            ']                               \n",
            "                                                                                                  \n",
            " block6d_bn (BatchNormalization  (None, None, None,   4608       ['block6d_dwconv[0][0]']         \n",
            " )                              1152)                                                             \n",
            "                                                                                                  \n",
            " block6d_activation (Activation  (None, None, None,   0          ['block6d_bn[0][0]']             \n",
            " )                              1152)                                                             \n",
            "                                                                                                  \n",
            " block6d_se_squeeze (GlobalAver  (None, 1152)        0           ['block6d_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block6d_se_reshape (Reshape)   (None, 1, 1, 1152)   0           ['block6d_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block6d_se_reduce (Conv2D)     (None, 1, 1, 48)     55344       ['block6d_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block6d_se_expand (Conv2D)     (None, 1, 1, 1152)   56448       ['block6d_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block6d_se_excite (Multiply)   (None, None, None,   0           ['block6d_activation[0][0]',     \n",
            "                                1152)                             'block6d_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block6d_project_conv (Conv2D)  (None, None, None,   221184      ['block6d_se_excite[0][0]']      \n",
            "                                192)                                                              \n",
            "                                                                                                  \n",
            " block6d_project_bn (BatchNorma  (None, None, None,   768        ['block6d_project_conv[0][0]']   \n",
            " lization)                      192)                                                              \n",
            "                                                                                                  \n",
            " block6d_drop (Dropout)         (None, None, None,   0           ['block6d_project_bn[0][0]']     \n",
            "                                192)                                                              \n",
            "                                                                                                  \n",
            " block6d_add (Add)              (None, None, None,   0           ['block6d_drop[0][0]',           \n",
            "                                192)                              'block6c_add[0][0]']            \n",
            "                                                                                                  \n",
            " block7a_expand_conv (Conv2D)   (None, None, None,   221184      ['block6d_add[0][0]']            \n",
            "                                1152)                                                             \n",
            "                                                                                                  \n",
            " block7a_expand_bn (BatchNormal  (None, None, None,   4608       ['block7a_expand_conv[0][0]']    \n",
            " ization)                       1152)                                                             \n",
            "                                                                                                  \n",
            " block7a_expand_activation (Act  (None, None, None,   0          ['block7a_expand_bn[0][0]']      \n",
            " ivation)                       1152)                                                             \n",
            "                                                                                                  \n",
            " block7a_dwconv (DepthwiseConv2  (None, None, None,   10368      ['block7a_expand_activation[0][0]\n",
            " D)                             1152)                            ']                               \n",
            "                                                                                                  \n",
            " block7a_bn (BatchNormalization  (None, None, None,   4608       ['block7a_dwconv[0][0]']         \n",
            " )                              1152)                                                             \n",
            "                                                                                                  \n",
            " block7a_activation (Activation  (None, None, None,   0          ['block7a_bn[0][0]']             \n",
            " )                              1152)                                                             \n",
            "                                                                                                  \n",
            " block7a_se_squeeze (GlobalAver  (None, 1152)        0           ['block7a_activation[0][0]']     \n",
            " agePooling2D)                                                                                    \n",
            "                                                                                                  \n",
            " block7a_se_reshape (Reshape)   (None, 1, 1, 1152)   0           ['block7a_se_squeeze[0][0]']     \n",
            "                                                                                                  \n",
            " block7a_se_reduce (Conv2D)     (None, 1, 1, 48)     55344       ['block7a_se_reshape[0][0]']     \n",
            "                                                                                                  \n",
            " block7a_se_expand (Conv2D)     (None, 1, 1, 1152)   56448       ['block7a_se_reduce[0][0]']      \n",
            "                                                                                                  \n",
            " block7a_se_excite (Multiply)   (None, None, None,   0           ['block7a_activation[0][0]',     \n",
            "                                1152)                             'block7a_se_expand[0][0]']      \n",
            "                                                                                                  \n",
            " block7a_project_conv (Conv2D)  (None, None, None,   368640      ['block7a_se_excite[0][0]']      \n",
            "                                320)                                                              \n",
            "                                                                                                  \n",
            " block7a_project_bn (BatchNorma  (None, None, None,   1280       ['block7a_project_conv[0][0]']   \n",
            " lization)                      320)                                                              \n",
            "                                                                                                  \n",
            " top_conv (Conv2D)              (None, None, None,   409600      ['block7a_project_bn[0][0]']     \n",
            "                                1280)                                                             \n",
            "                                                                                                  \n",
            " top_bn (BatchNormalization)    (None, None, None,   5120        ['top_conv[0][0]']               \n",
            "                                1280)                                                             \n",
            "                                                                                                  \n",
            " top_activation (Activation)    (None, None, None,   0           ['top_bn[0][0]']                 \n",
            "                                1280)                                                             \n",
            "                                                                                                  \n",
            "==================================================================================================\n",
            "Total params: 4,049,571\n",
            "Trainable params: 0\n",
            "Non-trainable params: 4,049,571\n",
            "__________________________________________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# How about a summary of our whole model?\n",
        "model_0.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hb-gspVenOiR",
        "outputId": "64654da8-7f74-472f-821b-7767b87adf07"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_layer (InputLayer)    [(None, 224, 224, 3)]     0         \n",
            "                                                                 \n",
            " efficientnetb0 (Functional)  (None, None, None, 1280)  4049571  \n",
            "                                                                 \n",
            " global_average_pooling_laye  (None, 1280)             0         \n",
            " r (GlobalAveragePooling2D)                                      \n",
            "                                                                 \n",
            " output_layer (Dense)        (None, 10)                12810     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 4,062,381\n",
            "Trainable params: 12,810\n",
            "Non-trainable params: 4,049,571\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Check out our model's training curves\n",
        "plot_loss_curves(history_0_percent)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "id": "9AE2jiRVnRe1",
        "outputId": "e7a2cb4b-bd97-484d-96f7-86b8a12ee2d2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Getting a feature vector from a trained model\n",
        "\n",
        "Let's demonstrate the Global Average Pooling 2D layer...\n",
        "\n",
        "We have a tensor after our model goes through `base_model` of shape (None, 7, 7, 1280).\n",
        "\n",
        "But then when it passes through GlobalAveragePooling2D, it turns into (None, 1280).\n",
        "\n",
        "Let's use a similar shaped tensor of (1, 4, 4, 3) and then pass it to GlobalAveragePooling2D."
      ],
      "metadata": {
        "id": "UiO6_bzSnjOh"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Define the input shape\n",
        "input_shape = (1, 4, 4, 3)\n",
        "\n",
        "# Create a random tensor\n",
        "tf.random.set_seed(42)\n",
        "input_tensor = tf.random.normal(input_shape)\n",
        "print(f\"Random input tensor:\\n {input_tensor}\\n\")\n",
        "\n",
        "# Pass the random tensor through a global average pooling 2D layer\n",
        "global_average_pooled_tensor = tf.keras.layers.GlobalAveragePooling2D()(input_tensor)\n",
        "print(f\"2D global average pooled random tensor:\\n {global_average_pooled_tensor}\\n\")\n",
        "\n",
        "# Check the shape of the different tensors\n",
        "print(f\"Shape of input tensor: {input_tensor.shape}\")\n",
        "print(f\"Shape of Global Average Pooled 2D tensor: {global_average_pooled_tensor.shape}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uLPSJ-h3or_M",
        "outputId": "a5605f44-d4d5-4cf1-a604-81259c901dc7"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Random input tensor:\n",
            " [[[[ 0.3274685  -0.8426258   0.3194337 ]\n",
            "   [-1.4075519  -2.3880599  -1.0392479 ]\n",
            "   [-0.5573232   0.539707    1.6994323 ]\n",
            "   [ 0.28893656 -1.5066116  -0.2645474 ]]\n",
            "\n",
            "  [[-0.59722406 -1.9171132  -0.62044144]\n",
            "   [ 0.8504023  -0.40604794 -3.0258412 ]\n",
            "   [ 0.9058464   0.29855987 -0.22561555]\n",
            "   [-0.7616443  -1.8917141  -0.93847126]]\n",
            "\n",
            "  [[ 0.77852213 -0.47338897  0.97772694]\n",
            "   [ 0.24694404  0.20573747 -0.5256233 ]\n",
            "   [ 0.32410017  0.02545409 -0.10638497]\n",
            "   [-0.6369475   1.1603122   0.2507359 ]]\n",
            "\n",
            "  [[-0.41728503  0.4012578  -1.4145443 ]\n",
            "   [-0.5931857  -1.6617213   0.33567193]\n",
            "   [ 0.10815629  0.23479682 -0.56668764]\n",
            "   [-0.35819843  0.88698614  0.52744764]]]]\n",
            "\n",
            "2D global average pooled random tensor:\n",
            " [[-0.09368646 -0.45840448 -0.2885598 ]]\n",
            "\n",
            "Shape of input tensor: (1, 4, 4, 3)\n",
            "Shape of Global Average Pooled 2D tensor: (1, 3)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Let's replicate the GlobalAveragePool2D layer\n",
        "tf.reduce_mean(input_tensor, axis=[1, 2])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fQpOskcUo8FV",
        "outputId": "8f1faf77-3214-4764-8158-e57429e1d4cc"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(1, 3), dtype=float32, numpy=array([[-0.09368646, -0.45840448, -0.2885598 ]], dtype=float32)>"
            ]
          },
          "metadata": {},
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "> **Practice:** Try to do the same with the above two cells but this time use `GlobalMaxPool2D`... and see what happens"
      ],
      "metadata": {
        "id": "A3QDe21EqlBr"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "> **Note:** One of the reasons feature extraction transfer learning is named how it is, is because what often happens is pretrained model outputs a **feature  vector** (a long tensor of numbers which represents the learned representation of the model on a particular sample, in our case, this is the output of the `tf.keras.layers.GlobalAveragePooling2D()` layer) which can then be used to extract patterns out of our own specific problem."
      ],
      "metadata": {
        "id": "dMhi_KFZsIrl"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Define the input shape\n",
        "input_shape = (1, 4, 4, 3)\n",
        "\n",
        "# Create a random tensor\n",
        "tf.random.set_seed(42)\n",
        "input_tensor = tf.random.normal(input_shape)\n",
        "print(f\"Random input tensor:\\n {input_tensor}\\n\")\n",
        "\n",
        "# Pass the random tensor through a GlobalMaxPool2D\n",
        "global_max_pooled_tensor = tf.keras.layers.GlobalMaxPool2D()(input_tensor)\n",
        "print(f\"2D GlobalMaxPool random tensor:\\n{global_max_pooled_tensor}\\n\")\n",
        "\n",
        "# Check the shapes of the different tensors\n",
        "print(f\"Shape of input tensor: {input_tensor.shape}\")\n",
        "print(f\"Shape of GlobalMaxPool2D tensor: {global_max_pooled_tensor.shape}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rAjwuql5rACA",
        "outputId": "a038f92a-6907-4c9a-f193-9aa610a25cf9"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Random input tensor:\n",
            " [[[[ 0.3274685  -0.8426258   0.3194337 ]\n",
            "   [-1.4075519  -2.3880599  -1.0392479 ]\n",
            "   [-0.5573232   0.539707    1.6994323 ]\n",
            "   [ 0.28893656 -1.5066116  -0.2645474 ]]\n",
            "\n",
            "  [[-0.59722406 -1.9171132  -0.62044144]\n",
            "   [ 0.8504023  -0.40604794 -3.0258412 ]\n",
            "   [ 0.9058464   0.29855987 -0.22561555]\n",
            "   [-0.7616443  -1.8917141  -0.93847126]]\n",
            "\n",
            "  [[ 0.77852213 -0.47338897  0.97772694]\n",
            "   [ 0.24694404  0.20573747 -0.5256233 ]\n",
            "   [ 0.32410017  0.02545409 -0.10638497]\n",
            "   [-0.6369475   1.1603122   0.2507359 ]]\n",
            "\n",
            "  [[-0.41728503  0.4012578  -1.4145443 ]\n",
            "   [-0.5931857  -1.6617213   0.33567193]\n",
            "   [ 0.10815629  0.23479682 -0.56668764]\n",
            "   [-0.35819843  0.88698614  0.52744764]]]]\n",
            "\n",
            "2D GlobalMaxPool random tensor:\n",
            "[[0.9058464 1.1603122 1.6994323]]\n",
            "\n",
            "Shape of input tensor: (1, 4, 4, 3)\n",
            "Shape of GlobalMaxPool2D tensor: (1, 3)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Let's replicate the GlobalMaxPool2D layer\n",
        "tf.reduce_max(input_tensor, axis=[1, 2])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QHBQn9mOrSIE",
        "outputId": "7b51491f-8d54-482d-85b2-169dcb38c2cf"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(1, 3), dtype=float32, numpy=array([[0.9058464, 1.1603122, 1.6994323]], dtype=float32)>"
            ]
          },
          "metadata": {},
          "execution_count": 20
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Running a series of transfer learning \n",
        "experiments\n",
        "\n",
        "We've seen the incredible results transfer learning can get with only 10% of the training data, but how does it go with 1% of the training data... how about we set up a bunck of experiments to find out:\n",
        "1. `model_1` - use feature extraction transfer learning with 1% of the training data with data augmentation\n",
        "2. `model_2` - use feature extraction transfer learning with 10% of the training with data augmentation\n",
        "3. `model_3` - use fine-tuning transfer learning on 10% of the trainig data with data augmentation\n",
        "4. `model_4` - use fine-tuning transfer learning on 100% of the training data with data augmentation\n",
        "\n",
        "> Note: throughout all experiments the same test dataset will be used to evaluate our model... this ensures consistency across evaluation metrics."
      ],
      "metadata": {
        "id": "NtQNCmvcsESt"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Getting and preprocessing data for model_1"
      ],
      "metadata": {
        "id": "vz0gZOrQvjdV"
      }
    },
    {
      "cell_type": "code",
      "source": [
        " # Download and unzip data - preprocessed from Food101\n",
        " !wget https://storage.googleapis.com/ztm_tf_course/food_vision/10_food_classes_1_percent.zip\n",
        " unzip_data(\"10_food_classes_1_percent.zip\") "
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "lLulcXNiujYc",
        "outputId": "9bdde806-c9e4-426b-81f3-ba1eafa48119"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "--2023-01-06 16:25:36--  https://storage.googleapis.com/ztm_tf_course/food_vision/10_food_classes_1_percent.zip\n",
            "Resolving storage.googleapis.com (storage.googleapis.com)... 173.194.214.128, 173.194.215.128, 173.194.216.128, ...\n",
            "Connecting to storage.googleapis.com (storage.googleapis.com)|173.194.214.128|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 133612354 (127M) [application/zip]\n",
            "Saving to: ‘10_food_classes_1_percent.zip’\n",
            "\n",
            "10_food_classes_1_p 100%[===================>] 127.42M   124MB/s    in 1.0s    \n",
            "\n",
            "2023-01-06 16:25:37 (124 MB/s) - ‘10_food_classes_1_percent.zip’ saved [133612354/133612354]\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Create training and test dirs\n",
        "train_dir_1_percent = \"10_food_classes_1_percent/train\"\n",
        "test_dir = \"10_food_classes_1_percent/test\""
      ],
      "metadata": {
        "id": "4WvWIss9vDvG"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# How many images are we working with?\n",
        "walk_through_dir(\"10_food_classes_1_percent\")"
      ],
      "metadata": {
        "id": "KqcPaWvdvOUg",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "29cea683-2b45-4490-ced4-9b203caafd93"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "There are 2 directories and 0 images in '10_food_classes_1_percent'.\n",
            "There are 10 directories and 0 images in '10_food_classes_1_percent/test'.\n",
            "There are 0 directories and 250 images in '10_food_classes_1_percent/test/chicken_wings'.\n",
            "There are 0 directories and 250 images in '10_food_classes_1_percent/test/steak'.\n",
            "There are 0 directories and 250 images in '10_food_classes_1_percent/test/hamburger'.\n",
            "There are 0 directories and 250 images in '10_food_classes_1_percent/test/sushi'.\n",
            "There are 0 directories and 250 images in '10_food_classes_1_percent/test/ice_cream'.\n",
            "There are 0 directories and 250 images in '10_food_classes_1_percent/test/grilled_salmon'.\n",
            "There are 0 directories and 250 images in '10_food_classes_1_percent/test/chicken_curry'.\n",
            "There are 0 directories and 250 images in '10_food_classes_1_percent/test/ramen'.\n",
            "There are 0 directories and 250 images in '10_food_classes_1_percent/test/pizza'.\n",
            "There are 0 directories and 250 images in '10_food_classes_1_percent/test/fried_rice'.\n",
            "There are 10 directories and 0 images in '10_food_classes_1_percent/train'.\n",
            "There are 0 directories and 7 images in '10_food_classes_1_percent/train/chicken_wings'.\n",
            "There are 0 directories and 7 images in '10_food_classes_1_percent/train/steak'.\n",
            "There are 0 directories and 7 images in '10_food_classes_1_percent/train/hamburger'.\n",
            "There are 0 directories and 7 images in '10_food_classes_1_percent/train/sushi'.\n",
            "There are 0 directories and 7 images in '10_food_classes_1_percent/train/ice_cream'.\n",
            "There are 0 directories and 7 images in '10_food_classes_1_percent/train/grilled_salmon'.\n",
            "There are 0 directories and 7 images in '10_food_classes_1_percent/train/chicken_curry'.\n",
            "There are 0 directories and 7 images in '10_food_classes_1_percent/train/ramen'.\n",
            "There are 0 directories and 7 images in '10_food_classes_1_percent/train/pizza'.\n",
            "There are 0 directories and 7 images in '10_food_classes_1_percent/train/fried_rice'.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Setup data loaders\n",
        "IMG_SIZE = (224, 224)\n",
        "BATCH_SIZE = 32\n",
        "train_data_1_percent = tf.keras.preprocessing.image_dataset_from_directory(train_dir_1_percent,\n",
        "                                                                           label_mode=\"categorical\",\n",
        "                                                                           image_size=IMG_SIZE,\n",
        "                                                                           batch_size=BATCH_SIZE) # default\n",
        "test_data = tf.keras.preprocessing.image_dataset_from_directory(test_dir,\n",
        "                                                                label_mode=\"categorical\",\n",
        "                                                                image_size=IMG_SIZE,\n",
        "                                                                batch_size=BATCH_SIZE)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "iTYQsbK7vVqV",
        "outputId": "7d235651-980c-49b7-dacb-3671f1395195"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Found 70 files belonging to 10 classes.\n",
            "Found 2500 files belonging to 10 classes.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Adding data augmentation right into the model\n",
        "\n",
        "To add data augmentation right into our models, we can use the layers inside:\n",
        "* `tf.keras.layers.experimental.preprocessing()` \n",
        "\n",
        "The benefits of using data augmentation inside the model are:\n",
        "* Preprocessing of images (augmenting them) runs on GPU (much faster)\n",
        "* Image data augmentation only happens during training, so we can still export our whole model and use it elsewhere."
      ],
      "metadata": {
        "id": "ZpLvz23TwHJ9"
      }
    },
    {
      "cell_type": "code",
      "source": [
        " import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras import layers\n",
        "from tensorflow.keras.layers.experimental import preprocessing\n",
        "\n",
        "# Create data augmentation stage with horizontal flipping, rotations, zooms, etc\n",
        "data_augmentation = keras.Sequential([\n",
        "    preprocessing.RandomFlip(\"horizontal\"),\n",
        "    preprocessing.RandomRotation(0.2),\n",
        "    preprocessing.RandomZoom(0.2),\n",
        "    preprocessing.RandomHeight(0.2),\n",
        "    preprocessing.RandomWidth(0.2),\n",
        "    # preprocessing.Rescale(1./255) # Keep for models like ResNet50V2 but EfficientNet already has rescaling layers\n",
        "], name=\"data_augmentation\")"
      ],
      "metadata": {
        "id": "O4GlzXR1zN5Q"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Visualize our data augmentation layer (and see whwat happnes to our data)"
      ],
      "metadata": {
        "id": "ITZD-qKj0Nrg"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# View a random image and compare it to its augmented version\n",
        "import matplotlib.pyplot as plt\n",
        "import matplotlib.image as mpimg\n",
        "import os\n",
        "import random\n",
        "target_class = random.choice(train_data_1_percent.class_names)\n",
        "target_dir = \"10_food_classes_1_percent/train/\" + target_class\n",
        "random_image = random.choice(os.listdir(target_dir))\n",
        "random_image_path = target_dir + \"/\" + random_image\n",
        "\n",
        "# Read in the random image\n",
        "img = mpimg.imread(random_image_path)\n",
        "plt.imshow(img)\n",
        "plt.title(f\"Original random image from class: {target_class}\")\n",
        "plt.axis(False)\n",
        "\n",
        "# Now let's plot our augmented random image\n",
        "augmented_img = data_augmentation(img, training=True)\n",
        "plt.figure()\n",
        "plt.axis(False)\n",
        "plt.title(f\"Augmented random image from class: {target_class}\")\n",
        "plt.imshow(augmented_img/255)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 529
        },
        "id": "5rxpYbLm2Qni",
        "outputId": "62141ee0-6e4a-453e-d71d-0eafef9fef2e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fbe031f9e50>"
            ]
          },
          "metadata": {},
          "execution_count": 26
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAASYAAAD3CAYAAACw0fnkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOy9efBlyVXf+TmZee999y2/vaq6qrqqF3W3toYGhCwhhAADBiEReAgbPAw7doBnwBCDBw+2x8GwOJgxjhl7xjBjHIMHY7PZA0YeFuOFAYFARpbUUrfULXV3de3Lb3/rvTczz/yR9/1+r0pdvUlI1RG/0/G63rtL3ryZJ795zveczJ+oKkdyJEdyJHeSmM90BY7kSI7kSG6VI2A6kiM5kjtOjoDpSI7kSO44OQKmIzmSI7nj5AiYjuRIjuSOkyNgOpIjOZI7Tv7UgElE/qaI/JNP9bUvoiwVkQc+FWW9jGd/m4i8+9PwnLMiMhIR+6f9rE9GRKQUkXeJyJ6I/Mpnuj5zEZEvEZGLn8Hn/66I/OVP07POiciXfzqe9akU92IuEpFvA34AeBWwD/wq8EOqunu7e1T1777YSryUa48EVPU80P9M1+NFyF8ATgDrquo/05U5kleOvKDFJCI/APxPwH8HLANvBu4BfkdE8tvc86IA7zMpkuTIlf3TlXuAJ28HSq8EPTmSQ/lUWugv2PeqetsPsASMgK+/5XgfuAF8R/v7h4F/Cfw8yaL6y+2xn1+451uAZ4Et4H8AzgFfvnD/z7ff7wUU+FbgPLAJ/K2Fcv4M8B5gF7gC/O9AvnBegQdu8z6/C/w48AfAFHgA+HbgI8AQeBr4roXrvwS4SLIWr7fP+/aF8+vAr7fv/F7gR4F3L5x/C/CfgL3237fcUpcfA/6wbeN3teX987a8/wTce5v3mLeRezllAf8AuNCeex/wRQvnSuD/BnbadvlB4OLC+VPAv2r7/xngr92mjv8jUANNW6fvBL6tbfv/pdWDHyNNdj/Xlvcs8LcB05axeP1u2z9vaY9faPvkW59Hf9eAnwUut+/za4v9unDdfw881erA48B/sXDuAeD/a/twE/il9ri09bretuOHgIefbzzd0vc/2r7bEPi3wMbC+V8BrrbP/D3g9Qvn/inwU8Bvtu36B8BdwP/avuNHgc9duP4c8EPte+207dFZaN9331K3g/HTPuungd8AxsCXA58HvL+t968AvwT82ML97wQ+0PbXHwKffUtd/gbwKFDR6u9zttELNOBXAf65CiAp7y8sAEsD/HmSFVZyM9i8rm3EtwI58JPt9c8HTD/TlvNI+xKvbc+/gWS1ufbajwDf/xKA6Tzw+vb+DHgHyUUV4IuBCfB5CwrsgR9pr/3q9vxqe/4XgV8GesDDwKV5R5MGxQ7wze2z/sv29/pCXT7ePnuZpDhPtp3vSIP1Z18CML3osoBvIgGXI4HuVQ6V9SdIA3EVuJukRBfbc4YEZH+n7cf7SWDxlbep50G/LgwED3xv++yyrdu/Bgbtez0JfOct1387YElAdh74R0AB/DnSAOnf5vn/L2ngrLb998W3Aaa/SAJcA3wDaRCebM/9AvC32nMd4K3t8a9s22KFpDuvXbjnG4FHXwCYngIeatvgd4GfWDj/HW17FCTA+cAtwLRJGgcd4D+QJohvWWij/3gLGHwYOEPSyT+gBRJeHDDtAV/Yvv8SafL4vrY9v440+czL+1wSUL+prcu3ts8vFurygbYu5fNizwsA0zcBV29z7ieA31lQwN+7nVKSFPkXFs512xd6PmC6e+H69wJ/6Tb1+H7gV18CMP3IC7zzrwHft6DAUxaAuW34N7cN3wCvWTj3dzkEpm8G3ntL2e8Bvm2hLouW4N8HfnPh99ewoJAvApheVlnt+R3gkfb7TUBDsn7nwPQm4Pwt9/4QtwfQg35dGAjnF37bVg9et3Dsu4DfXbj+YwvnPqt97xMLx7aAz3mOZ58EIu0kcsu5L2EBmJ7j/AeAr22//xzwjxf1sT3+Z0kg+mZaC+/Fftr++tsLv/9r4Lduc+1K+87L7e9/CvzMwvnvBT5ySxvtLvw+B3z3wu+vBp5aaN8XAqafWzj3NtLkKwvH3s0hMP008KO3lPcEhxPCOVov64U+L8SxbAIbt/EHT7bn53Lheco5tXheVSckhXo+ubrwfUJL9orIQyLyb0Tkqojsk8Bg4wXKWpSb6ikibxeRPxKRbRHZJXXcYnlbejNHMq/LMdKsv1jeswvfT93ye37+9MLvawvfp8/x+6UQ3C+6LBH56yLykTZatkuysubvfFNf3fL9HuCUiOzOP8DfJBHcL1YWy9sgzbyL7fRCbYSqvph2OgNsq+rOC1VIRL5FRD6w8E4Pc9geP0iyiN4rIo+JyHe0dfgPJBrhHwHXReQfi8jSCz1rQW6n31ZEfkJEnmr1+1x7zaJOvlS9uVVHT72Eei7eewq4pC3KPMf5e4AfuEU/ztzyvOfDiQN5IWB6D8mN+rrFgyLSB94O/PuFw4uVvVWukNyC+f0lyZV4OfLTJD/6QVVdIg0MeQn3H9RTRAoSX/KTpFl4heRPv5jybpDcjDMLx84ufL9M6ihuOX/pJdT1Uy4i8kWkwfb1JGtihWSuz9/5pr7i5ve7ADyjqisLn4GqfvVLqMKinmySrM7FdvpUtdEFYE1EVp7vIhG5h0QbfA/JzV4huT4CoKpXVfWvqOopkjX3U/N0FFX9h6r6BhJV8RApQPTJyjcCX0tyw5dJ1jG8NB2/VW7V0cvt9zHJe0kPELnrOe5d7K8rwGkRWazLrfrx47foR1dVf+E25d1WnheYVHWPRGL+byLyVSKSici9JF7lIvDPXsxDSMT414jIW9pI3g/z8ht6QCIbRyLyGuCvvsxyIPEkBS3IiMjbSbzFC4qqBuD/AX5YRLoi8jqSTz2X3wAeEpFvFBEnIt9AUuB/80nU91MhAxKg3gCciPwdEncwl18GfkhEVkXkNGnAzuW9wFBE/kabo2RF5GEReePLqUjbhr8M/LiIDFqQ+G9JQZRPSlT1Cokg/qn2XTIRedtzXNojDZYbACLy7SSLifb3XxSROVDvtNdGEXmjiLxJRDLSAJ+RXMdPVgYkY2CLBBqfilSa/0ZE7haRNRJf9kvt8Q8CrxeRzxGRDmlcPp+8BwjA97Q6/bWkYNRcfgb47rZdRER6IvIOERm81Aq/YLhcVf9nklXykyRA+GMSMn6ZqlYv5iGq+hjJF/5FEuqOSFzNi7r/FvnrpFllSGqIX3r+y5+3XkPgr5EGx05b7q+/hCK+h2Q2XyX54z+7UPYWKULxAyQl+0Hgnaq6+YnFfFrlt4HfIvEjz5IG1KJ5/SOkSecZ4N+RJpUKDoDkncDntOc3gX9CmtlfrnwvaWA/TeIr/gXwf30S5S3KN5Msso+S9O37b71AVR8ncXLvIblEn0UiiOfyRuCPRWRE0o3vU9WnSWD+MyS9mUeb/x6AiPxXIvLYy6zzz7XlXSIFMf7oZZazKP+CFPl7mkS6/xiAqj5J6u9/B3yM1P63FVWtSd7Td5Kibt9Emmjn+vEnwF8hubg7pIDMt72cCsvN7uKnR1pXcJfkjj3zaa/AkbxoEZG/Sgo8fPFnui5HcueJiPwx8H+o6s++4MUvQT5tCYYi8jWty9MjWV8f4pDYO5I7RETkpIh8oYgYEXk1yeL71c90vY7kzhAR+WIRuat15b4V+GySBf4plU9n5u3XkjgpAf6ENAt/+s21I3khyYH/E7iPZNX+Iimh70iOBODVHObuPQ38hZbP+5TKZ8SVO5IjOZIjeT45Wit2JEdyJHec3LGLKB8w8tym3AsaePKcv17ibZ86UbmpcC9KNIpFMGowakAtimAICEoEVLT9RBDFBAFtj7UlihhEBXMbqzcieGMwku416Dw5B9H2eyoJBRRBsFgUg0eIqT6SYsReBC/plRRFI1gMJloEQ22rVDcFUQNYROdPkXSPQJT0frF9ulHSexIQwEbBqSAYDufOiAgoEVVta7zQzC+jaz6hy2+jA40IXgyoaS9SEA8yb8/bPEAXC53fK6C3sQgEDEIWhZQqFDFACobGth8Onzkv0YqgelAzlHSd0qofqU9UDCwv84u/8S4e+TNvwhiLuzkn6Y6ROxaYbif6PM0oBxn1L+Pel12jF5CkEQdPsKrYqIgkoFA8iLRKlUBgXh+RlBijEVRbMLjpXdLQNgvvnI5Lq6WKaEqtiZqAwLTlpqGdlJ+FOUBMQ4w5EnOEgBBQ1RYyBaPhAGDSY2I7QIUsHJTSDsHYfmM+nNIyBW2Bt4VY0QTHKqlutl2ScAiX6d3R+bDTg+d8QnPf8lu5PYB9wt23URBRcDc9V28u4HlRcT6N3HzRrcA6vzS1V9IM4WBZB2AO2iVdenh/bK/RWz6vZHnFARM8T6NLmn2f8x65/X1/6lOGHKqKaWc6VYjzgTmvhGYJJ1qlTDbHvNY2laER5laVmoOhf6C+0o7u+V0xwZYhEucWl4JiUDUgEdG5NRIgJutITEQ0zeCCYFURMViUEANeQU0CQDVKVMVqGjbz+i+2eDywz3SheoegnazD+Uu0azhRhNjWOX3m996uz3Th/weP+ITzC11z2x83H3aq3Jw/eVjwC4PAvC3Su95sQ8PcaFFNbRAkou3kdQjrkkB73rIKKskK9brYpguPvCNtoRcnrwhgWvRUpHUlUBAzRxud/3NTJy9aqcKh4ktrUYhIui4uWAwL99xaxq2BgttbwUlR4nNNXSapd1BD69AhIlggw2EkYm1DkUGngDwzWBFMa+Y76xADPkaCCFHSuxhjDkz3OQiHAM1UmM4qMEpQxVjIihzB0FQNdRXaBjYY2u12TACJFM7ijCARLJYYI9UMam9oxNKIoeiVbBxfwzhzYNAk4I1ojMQYUTS5IK1laKS1gWICm6hgrUGsIEYw1qIKxjpCjIdt3/or8+fMrbHDfljox7nViBwuDl3ot6iKxnioB60O3KQzIhhjMFGRGBeAcQFIWnya69Lis5XW1BVBNR4CmWrbBumeGA9MTUKICY41sr+3RzWr2hLT24aoc28wfYxJ5RmTLCdN7UrUg3vii0fQO0ZeEcA0lzQGhY5zaSC2nS4HypesCeZuj3JwXDTZHMYYrLFpkBg5cC2Swi+4RC3SPRf4vOCxltNpmoYQbp7NEn8kqBhEwEkk00g3M5zeMNxzZpm1tYxjxwuyrKLXtRSFozAWK0IndyiRSV3TIKi1ZLmj2+uBCLOmpgmeWVMzmdQ0TU5ZloQYMM7gfcPGxjHQyHg0QqNhZ3Of/b0Jk+GMyXjGdNpQdgusdeSug4mGTpajTeDj54acv7zHLGaUqyf49u/+Lt76pW9FMlBnaLyHqNR1xXQ8ZTqZ0PiGJgZcZul2uwz6A5qmYjjcZzga0zQNS4MeZZGnKUYMeadDtz+g7HVpGg8ITd0wm82IMQGptUkPjDFojIhJIB9CIMaY+tqmvvbet/2qB79nsxkxBKy1SY9ED8BoPtiNMeTWQgyE4A9ARKMQ51gjJABuJ8Y5BM5dsaapaXxD8B4RCDGACiEGYghUVY2qUlWzBOxRmUymvOvX38UzTz2FMQ4rqS6i6b2yIqfodCi7Xfr9PmXe4dKFy1y/ep3QBOrZjNDENDvK3B1+5cgrApjacY6IkGcZhcvxwS9YUtqaJxwQs/P7RIQYkxkscxu4VWKjySyOLYile2Rh9hNutZLgEy2p+TFjDilNMZY8tzRNk8AxpgoZTdaEcYqVwF1rjgfv7vPAmWVOrEX6faU/CIR4hTwPlGVGDCFZEskBY3l5GVNk7M0q1BpOnTrF2sYy09mUy9e2WD+2gZo+dRNwrkue59y4fp1Op0Do0OsZggqVH9Atu1y6UFFNlV65znQ8xtdCaCybW0N2dsZYcnpFRjWumTaR3XGgGU159evO8BXv+FI6Sz1MIdiyQwiB4CPBR6pZ3X73jKcjQCm7Xbq9Lja37Ozu4i9eJAz3yQYd8jwDTQM8qGcWKvrdZbr5oO0TGA5HzGYViMW5DOcczlmaJoGGcy5xYiEmK6ztkxiTe2Rba0SjUlQVjW9aABLEJGvDWJOsDgGNSqfIKfKMpqmIMRKjJjdYLYGAWDmYJLUFxPSxGGMOAM269B1orUUOzifLqQ1K+IYQAhtnjzMejTFYyrKk2+uSZRlZllF2OpRlSVEUlJ2SwhT80e+/h7//936SelJRTQqmwwl11SQeEA6ox1eC3LnAtOAjt31O5hzOWjR6JIZkKc3dsAUAmUeA5qSjFXNo/0cQkxSAGDEihLnlJfNoiByCIYez4vwZt7OYTDtjxxgxkmDESGtiExOwmPRyG6uW1716ndfc32O9O2bQGdLvWIwJONewsloyWO6yt7eHtTnD2YTgG1ZWBhw/sYQXYSVfpVZlf3eL6XSPrMg4c/cGVWg4d+E8vcGAfq/H7k5NPRkx3K4wIsT1Nfqra/SWeozH29hiwqnjPcrcMp0Ik9EUZ0tO3rcOehqjOX7mGe8N2TiVc+LsiD983znuua/LeHIBOms4V+CaLtbl5J0MNCMvOxhjaeqGzignaABrsJ2coiw4VnaQzHLp/Dl8U1P7iGsBqK6SlbWzlXH69Gmssagq/V4XY4RZHVKET1t+y9qDiSHGcMBozamgLMuo6xrvfQIvSROJlQRCIQayzBA1trNbbPUBQozUwWMyi4nSupEWIw4yg3OWvCiwNllo8z43cqiH3jeISW6stYYDrsmYVl+kvTJiNOB94L5Xv6q17i3ESDSC14CGQ/6PlhB3MeMNb/p8zt53D49/8DGKIqewOft7+0yqpnUvXymwdCcD04LxKS2/khmLxEikRiVpnDHz0FXiWKIeMA+HLl7U5Ge33IYR087AEWlNZF3gIow5pCeNQmgjRvOo0YHlNee6xBzwJlEDUQNWA2gfR0nQBidTlvqeTqbcd3adz/3sVTZWpmRs4uwM5xrEltjMsbS8RG/QweaGcqVPFCU2Q06cXkWl4vLoHFme089WqaqIdYblpZygNYRdQlNx7FgHV1rEetY6HY4PBlw8dxnX6WAyx/WdLU70S7pLfYIGrm3eYDrcp5vnqGvorzjEGDLjcCIYF6jqGafu8ayeFB74rLMMNhq2b/waYs/CuIPYVfqDYxS9Y5h8BZN1UVuQFRlic+oQkSzDZBlqLFYNK8srmNNn2NvbYTYd47WNNOYZRaeHqCPMAq5jcVmGiqGwjjCpaOomcS5NCqtLa4EYXOK4AklPjBAiiHGQWSIkAMJgxCXuRgMiQqcoiBqJEtvpLQFMYuAsxhlc5siLgizLsFkGJk1KIQZ8kywi27qDcx101i3oomCMTaWHpD+HPFUby7QW1ZQigEQiAZWIGkVFUjpGJE2wqkQXWNoY8PZ3fCVPPPYYs8mYbl5w6u51Ll2qqWdNeoaCqCwEJ+5MsLpzgekwVJV+GiG2EaWoBWjAmNSRQTXNXhoxEg8UAkizjdFERLYzRpiDlpiU2yGRFt1arzCQrCZJUau5d75gCi9yCRCTab8QeWo0IFrhTCC3Mx58VZ+3feFDOB3SzwJlMUL8mLWlJZo6p1s6Nk6ukheOzc2rXLhwkd5SiVrwGjlx92nWjy3hw4jdvRsA9Je62GnN3u4OjSq9ssd4MqOJlqJcx0qHKxeuI80Q0zQYgcloj539bchLZoOaopOzNlhnvLXP1c0xO/U+WZkhrNEpS85dvMp4f8apu47jxDEeXsDHQJF3iH7CcPsJJuOcrCjw4SRF5xhZfhd5eYaltftZWr2bvFxCM7Aux7kSIZHq6itM1iFbP0O3XOPGjauMhsPWKxd88Bi/x2gsqA7omiUcll5WYLs5+36PpmkQSRYTgIbQEuqKdS4FSKwhtOR1jBFrk/WViHZDDOGAjLbWYtQQ8EC6XoAiy+j1+2RZlmgAaxNHKdJOWEmnFvmpGA+jeJ/o/s+vCYhJEdd5wGB+fYyKsXO2XxJdJIb0NzQMgWSZ06Z1QOTNX/C53P+q08yGEz73kUe46/gJfv1f/yZPfuSpxMPdNMjuXNbpjgUmXbB82lAOHp9mEJOjatpIW4pioAHbRrdCa80sxmqCpiiJkmYzaw0a20ygxWRDA9pGkgwtGMo856dNZGvNfMS2Ye4U5bFRMGJRLEEDThoyO+O+Mx2+5qvO8JqHclywnH/qHLkTRtPA5o0K30SWX3U3Fy5cJsuhUzqMcfigYBxn770PKR2jekaWWbpLA0QiKp7GTylKi2pgc2eHq9d2WV45znpnjTBVzhw7w3S0xZXzH6fsZBSuoKqUfr/P5sUrWCecOnWcpU7BidUB+MC1GxOe+dD5lKAZA85ZPvaRpyhLx9KgwTcNRWZwdSDEiq7NsRH2p5fZvt5haekezm/9CU98bI/VtQc4de8DnPmc17K2foa1tXuxtgcRvCquyKm1ptaacqnHjd1tmtrjXI4LAYtj1gSY1ShTxDq6PUe37BCDZ3dnF980NLGiUxTpewgYa8gErHEJrFpdcM5hjSG0ZLiPnhhSblaZF4Q5sMWI9w3WWpxzZFlG3oISIsklM5ZG402gM3fp5+UkXdaDa+bu/vy4mDZCqTER6wpwCG60IBtjwDiLWEtoEg9nxGCSwiPRY8TT7Vm+/hu+hjCryWxGPavpD9KWU6+kPwp0xwJTEtv6xhGbCd3VDJtbbNbB2pSU51ziFgSHtS6Z6+0koyQiE4QYPAaIQYkh4muPkPigZhYIvk0kDJ6ogeATYRlDgigwiURv3TWjIJKBAeMS1xU8CBkoZDbimHB8zfCWz7+L1z+4zPrKlM1rW5y5Z4lOp+TcuU0yV5LnHa4NL1MWBWW/z3C4R9krGVczmqoh39yF0rI33Ma5GRsbXTbWl5nWFddu7NBxJWfPHqPsKtOZEGPDZLSdQs1VhWhFv6fkueXG1pDxGDrFhPFoh6CBwo6BBhOHZM5xej3HuRwxgssdQVs+B6VqZmjMICjNeEont/RihhXIOlNq2zAZPknWFBwrPbvXrnB++EEuXejzuoffTPnIFzPYeBVT7yDrINJnMtlnMhuzsrLMxvENrly+TlDoZH0wjio4Yq2QRTLTkFHRyzqsLq9QZDn7+/tU0xnGWnJrca0tG2M8IKSdc4QQDjlAYzC5QwmYwh0EVlJUzxNC4pfyoqAoCsQculsIBI1oBLGJpxIRmqahaRoAQkjRuzkRvghK8+DK3E1MkT5to4vznKXkyqWosbT0ZsqGt20KhxGF4NHgqSZDrl+6wLWLF7BSI7ZBo8daDhjXFiY/zWP45ckdC0yHWRhJQV79+gf483/pK3AldLpd8sxhHCCKy3JUBZuVIEUKDUsKyxtrqX2DbZdPNHWDr5qDqJFGpZ5BNaupm4qqmjEaD5nNpkwnY/b3pwyHE8aThum4YjaeEOoZvq4IXggEGgJqwGWO6BOcWm1YG8CXfOH9PPzaVerxNo8+fY611SWOn1ynYcap++/izJlXsb+3z2jUY31pnac//jS2yOgMlhjtBbrdLvc++BqiBKKe4vrmOTaOLyMmcPn8ZWyWk7se2zsj8rLL2sZxmqZie/MG4+E+G4MOKysDdrdmB6R8t8zYWO1T5hXD6R6+3qYoDEsDUO+xHUMMM8pOjnOWOlSoCE2MDJZbtygIGhyhCURGZDYjN4Y8c+SZ0isDx1Y6NLUjRsHLGLv/Pj723mcw3btYu/u1dFZP0186TbNf00HoqOW+EwNWO4YbN3awNmCE5HITUyRWIKsN3RLEWXpLAzrdkuAD3nuGwyGzuqKua0IImDbqpi1B7oxrJzQHRDpl3nKNyQICEOlgjBCiPyDIAYKAWDlIDXBtQMQag/eeyWRC0zTJ/ZPkFpZliXPuAJAO3DmZZ3UfWjKq8SAS2F6UuCiNqFo0Kk1dIRGyNhO8noy4fuUyF849zWy8jwkNmURMZggegkjiTA/YiDuTU7pV7lhgSnS3aYNhysnT63zB2z4PW3qij20KdeKOQozJ/8ZhyJNrNZ8lDG0oGSS2M5IaDIboW1JSMkKMNE2dyOvoU7ne0zSGahaZTSP7eyMmwxH1dMzO5ia7m0O29jbZ2ttkuD9ktl/DTJCqYb00fNmXvp4v+sJTbF79MJsXNzEUrAxW+PAHz7N0zHHyzAaXr1xEg2DU8fgTH2c0mvDa1z3A8rFVrn5wl9FwxIc/+iTL3RJlyumzJ+jkhr3hDkjOibtOsTFYovbKox9+kqoKGLEYDQy6y2jw1DNLjF3UWwZLBUEjs9keop5e6Sh7BjERrykhUjJDZgGpqIJPPIexFM5RhxrVRLjjHGIjxgpqGqbTCJoiQCFMmI2V6B3W5HTLgjDdJtbn8NMuT1x8N6O6Q97dwNgSl2UYm7G8tMbq+gnWij7GrhJ0gJo+mCWMZoh3UAnBRzBgrcXmOTZTnEbILG48YTKZMJvN0Bjx3uO9x1rLTKqUQW0NYlKENn23aBvdEFUy4xI/KYlHQiQR8zESDoDMpQhcjAcRvxgjTZNcwCzLDvKoPhGY5qks/qbjUTRZ/QDMM97TtUE9BsUZgx9PuHLxIpefeZrJ/h4oOCcUWY42kRAa6mnNbNoQfLIUrZhXiL10RwPTzaLA/mifpbLAZYJXT9AGsUqWZQSvEAMxVimrF0VsWnNkXcTiEAzqk/lrjUMdxFARjCdaxWYRQ6BpKqRdquFiQWeQ0/OOlWN90IiVQDUZE2tDE2p29reZjEbsXd9m+/INti9d4g0PHOOh+/s88+T7sXGfTCz7ezUf2/s4x46XTPdnfOzxZxMZKzk+RCKRh17zIMdP383eeB+1joc/67OZVDMuPvlxMhMoc0MVh+yPh6gql85fZb+7yfLqOssrA7J8wPr6CUQjOzeusXf1OheevUbuCpZWunjG+Dglz6HbyckLS5bZgyS/FGSIqEmheJc7XOaYVU1y17IcVGh8IM8LMA1iUhJfVrTRzCjUs5Q8KOrIXUZmI3kWCGFC8GNWxZEFodm7SDnoUNgMBNwkY3tP8Y2g9iQ2P0vWuYtjp15HZ/ksYlcwmqOhAcloQspBslbQoJSdkk7eYbA0YDQZUlUVvm5o6kAMaaGw9wEfEreT6h7JiwJjHJnLACXWIVlLLllRtk3cDLHlKcX8zbQAACAASURBVG1KLTBi8d5TVTWQXDfnHMZaOp1Oa63FQ14J2mRQaSOJYe4XpOtIudpi2nwrDYgBZwTTRGaTCVeuXOfaxUvs37hBFiPG1wkgp55aAvVsjG88vpaDqGNa7xjnGRSkFNU713q6g4FpvlbKIGL5+Mee5Z//s3/Fl37lF/Dwww+Q5Xkyo51BfcSJIYpB3TxJL0Ab9VBjcGKQEEm2VEhpB5qiHnae4iFK4z3GRZCI93Ub/euQ5QbrDDFENNZkPUW70JGC3vIaTgeEM11mD3Q5/9F9Vtnm8tNP0jMVG0srTMYjikHG2kqPk2fWmWrg4qVLnD17hsHKCh/9+JMpzQDD448+ztZwh+Vjaxw/fgzjHLZquH7hHBefusTG6RU21o4zG8/Y395jtDfhxuaE9eN3UXaX6PQGVPWQcRih1rKzC6958B7W1guuXH+Sbi8nLzzOVvgAs2kihJMpodjMEtsophhD4wPGCp3C0cRICDElJIpiRMkzl+pepKgRwSK9jMn+DI1pqUrwE9LSOoONitWAmIAYRxZqpJqA8/TyHNuxFP2CvfrjeLnCZL/k3PYf0j/2IMfv/Ty6y/cj8VWIrtFEh5eItRFLW75JlootLCF6Qu3RBpq6oQoTmtq3vE2kbmZg0mSkoaGSKd4nF65oUwJcZnEuLR+CZAERI41AEJ8SLucckSrWWay1hBCYzaZpCZGv8aEGQsuJCoSIa/ORrEvWmc1yxGYYZ7BW8LGhrqZU0wmTvT22r1xhf3sXP63oGoMRpTEBP92nrqaJ2A+BbtlnGhvAIO06xEi7VEpgvgryTpU7FpgOF4KCRrh84SrDaou19RXuPXOKpdUBmU3RlRikzWEpCCEjswl+5uY7gKEh6CwlOKoSo8dmliZ4VKX9pEQ3MRlBPVlWkhuHDxbvJeVKRYMzBQHBS0NmlNyBjve5cukx9q6exw438X7GatdRGmhG+xTi6PaXOLa2xtb1LbZGe4gxrHQHbF66Sr0zogk154Z7BIkU/S7V9h47Fy9hnGNn6xLKlAceOMXKiWVM4agmNZe8R0zGZFpx/eI2Ny7PcK8tWdnI6JY5e6bm7rPrnL3nFNPJNhoDuXOtS5IUNXglSkPmLNalbOZ5hrOxKXHLSAqtpy1KhGk9YzYLWKspcdBAXjgMlugN0nX0+5Grl6ZYpxQlpLSKiDFpSUc3y4gBfKiIQakbTwCKwlKulqyWDeLGTKsZVRgxCXs8+/iT1GGNnfH93PvgW7j3oUcYrK+iookjCgYTOu3ylBxLjhQKuaK9SNCVlGmNJ8aG4BtA8T7Q1KG1fqqWRxKm0wqp0vtlrsAY27pn0DQ10ibMhpAAO4FwIIRIXVd43xCiByIhNknvrMGpRUOgsBlOEkcl0kbajEkpKwam0xHj4T7qG/AN4hs6VqiI+NmEajJCQ8ruti71lXOOoB6bSbIS2wi1pKq+IuQOBqb5Jh2JBo81THY9//G3/oBmNuLt73w7K+vL5J088SE25RKZkGZNZww2z6kbj7EWCLh2NjFZitSoiYhEbEy7EIUIGIsPAUKKuITgQB2oQdSSZ2k7ECMGIx6pp4TJLuPLT3HtI++j1AnHC4uxNYV4NMQUjhZhb3/E9a1NYqa4fodQNTz66IcR7+kWObFQllZ6qDXsTybs7uzywfe9n+MnN4g6pdvLKDqGvb0dGpRmVnNsY4OyWCLLulQzw8WLN+i5LqEaY03KpwqqnDv/DKO9axSdiJ9VLfDaZAlaxWYOa9sIkZ0HHZIl4DKLiFI3Fd4b6tpTt66OMckx6Pd7VNUYFSEExUhGlgtr6ym8b2yDMYL3kW5Z4ptIDAkQUn6NI48F40mNDwHvb9DtFSgOMZFuP6dAybXh4qWLPP6fP8Tv//avs3Hyft741rfyukc+i9W1Y3R7q2C6iBZYSoScoKZ1YQRr85R60q55s86BCtYI3bLNhWrD+yGENtJWU1UzxuMpqsmNc64l5P0kJWQGpWkCogbvAzCPxNEulTFoSPlIprW9VCH6QBMaovf4qkE0EuuG2lftrgoeS8RoJDYVvpkxHo7TesTpFEPASiBi2nwok9YSisNlGV601VUOdreRhc+dKncwMN38yxlHgWW6U/Fv3/V7PPGhp7n/wfv4gre+kTe8+RG8TMGkBDJjpeUChMxJG/6PiKQ8EASMsW1ym0OwaDA4Y/E+kEuGRWmqgG8y6jq0UboJs+mIejqkyKCe3WDn8kVWnFJde5Z8WrHeL9BmimQguSV6KMo+k0lgMqmZ1A3WFLz+wVczmQyZ7mxhm5rVjTVWjy9R+4Zrm5t0l1ZYPXmS4Wyfh177atRMeewD7+fcs5dYPrbG5RubaFDOnsipJ4KVGbvbEy5euM5stke+XNMwISv7XDh3g81ml+NrOWWRoWGa3AdjMZIy3V1m8L5O7i0uRaWCtjOwQTXQ+AYxOTEGstzS73domgprFR+q5Ao3DU1tGO7tQBTKrmCdx+aCswbjAz7ODrY5UUk8jzUGXytl6RAJxBiYzgzGdYkoGSV50eX6tT3qcc3bPv8uGu/Y2R+y+cRv8r6rf0DZX2bQ38AtL7O6fopjdz1If+UMJlsmUmBMBtHj2qTLaEs0xLQ1Xpt3lPKHLAI4l1EUnZTT1Hj6vYYYlelsQowNja+w1hJ9SOkBmqx0g2mXrbSZ1iSL3pm0a0JsIpi0CBkfGQ/H+FmF1j7t1wWgHteurRP1jEf7+KaC6A92DyiyDCHRCyFYYsxBW1fWJS5QJFl+C/kCbZ3ubLnDgandzkzBIax0+7gC9kcV5z96iWvnr/PRRx+nab6OR974EP3lHo0A+LQMQdMKfisGm3UTCWig8jWqbShaIz7AZDSmnjbMJjXT4ZStzR2GeyN298bsbG8xnQyZjnYwOqV0kdxGurZhpZMWV9rplJ7tEOpIiJaik4PJGVcjRpOK4X5FDJB3S46fPM29D72O4XCL9/7eRbpWqXzDpSvbXNva4+y993D32bvZHW9zY28Hm+WYLCcvVji2vs7g2CrB9djZ3OZjT13Cxda2bCKhUW7cmLAsXU7ff5y9UUN/uYNOPUWRQawQhbX1JbJuzniyi5iYFpca2mUcieHLiizxaoTE2Um7T5RVemWHGD1FJy1WhniQLjHcr2hqpezkZE4wNu0f5WODMUqnU7Shb0+hUFUelYhYDhJfNQhVaOh0hTNn7mMyGxElcOxkH+cm9MrA8WMn0Jhx8eIlLl3+GMPzNUM6uEGXre6AZzsrdJfvZuP0a1g/+SBrJ+7FmgFKCdJDpYMaQdSn1JM2sXKegxTbbHHE4FwOpEXZZdklxhpTw2TqicG0OW+kdWxGDtZgpgiw4BuPMWkFg0haXBxDZDwa01Q1JijapIhzILnHs+mUGCtCM6Oppqlz0JTEG9MyGmskWUs2rU80JsfaPF0jmoC4pTNeSXIHA9Pheh5pB0SvLDFGmcSCuq5pQsO12TX+5S/8Il6+gre87S1YMyACVixEIcTkLoRGmVWe2tfsjvcYjYeM9vfY29lmuD9ib2eP2agmzALSGHwVCJXHN2OcVXLnWWFMr6hZKaGwaauSQmpc5ekVORpgVgfyoofRgvH+jMlIybOcwlk63ZJOt4N6uHDuIlevX8A3AbGW889eQKVkdeMUJ4/dx8XzV3j20lOIq9m6vs3+rGJ7t6HswDQOOXnPAxw/XvHEox/CBU9mPMQGEYfkHdZPH2P1+BrlUsPe5pRJtY8RJYZA4YROnhMlkBcGNZFATK6AckDgJjK+3SomzxJ4qSGqxZiUaJrI4TSIg09JqM61Lq8qxiRuRUxKcMQqdV3jMpsSZ8XSyV3ak8aCsYZqoiAFMSqzyRiCodtZ5dkLTzHodslcw+Wtp9ibXOX0ibOcvCvtuGlO9JkMPftbO4y3z6NlRth5nPGND3Dx4ydZOf4A9732z7J24jUY6eJx7bq1iGoixIE2xylF7dIWKu3ylZQ4hzaJNrDGYk0GWUrw9VqDWELjWxDOEEllIFBXNVnrFkej1E1D7ZsFDijlM/kQGI9GBK2JoQZNrl7uLM5maWcCSftYNU1NXuSIWiQmt5Ho04LpNsE4ba2cFhA/x2YZd6TcscBkcQR8iqih9LtKtzNle3tIrKc4Y8hd6qTJ9RHnHvsIr7rvbtaOP8hoCHvDiq3dTba3txjuTdjbGTGbTajqEUEDjffUTY3LkoJNRhOsGnp5SafjMF3LZL+iGo8wYcxGYVnKI8ulUDiPNYGOdKhngW5ngHohik08jUZm4xlV3eCkoFv0sF0oywLnDKPRPk++//0E43no1Q+yu3UVzZQyX0bw/Of3vYeqmRFNoCw7XLh0neFoyGxac+nqHp//pjfQKzuMwwyVKaqepqkpcoBAXXmuXkxb9nb6HY6tLXNuc5c6BGxmCTGtrwvWk3UMkkWiSSFxoiaOJ4C1EI2CVZo6WTvGOrLMpHMkKyqqwXvFN1A3EZdL2tJXI0KgLBzGOmazgIihaQIhpI3agijuIKcINDRpmYWNlGoJk4YnP/xhbN4hL3PG4wnLq0usnFxhPBoz9UMoOlSxop7OMLag7ECWdbCF4cyDJ6Ho8LFnn+TZjzzJRx77AK9/5M/x6kfeQV6eJUqJKRJZLFJhcEhodwggYh04SfxQVKUsO3S7nXarksCsrplOpml3hOmUajbBIzR1yvfyTdodVIykXQ1QMpclvicqRdHBxwqJAZPBZDamqmbtkpa00VyIkcxlaYFvk7ZTcUYI2iSr1rZ7fQaDc512tXkKsmo0aEy8VjxITDj07O5Ul+6OBaa0yFaJ7Tq0jfWCu050GO5vsTRwSO7Isw71WDGVZ//6mN/5jX/P/uz32dqKzLzB6xRnA4Xr4NTibMDZGaIeaSJaV8ykweQdwFH2lljqdpDZhJ2rz0AzJaOmcJ71bocSYbXTP9wmwzpsltMtlhnVU7Isx1cp8tf4QPSwsrKcFpk6xeWQZYpOZuRWWD5xnPtf/RBbO8s8/tEPM5OK0IypwpSVtSVO3X2au+4+BU7Y3trk0UcfJ8tyRA2xatjfuUG3NAgZs/EML4KzSlnkXLm+x3B/yInTy5TdZYquZVRVOJsBgm0i1hiaOuCsAZN2T3TWYMWikhL2lCblI1ltB1hMbg+aQtoZbQSq3aqkUbKsSBvkRyiyHGLAOCXPLCGAcYlrURNxuSVKbDfUD0T1iGnD5+oZmESITGZTNtZXmdYztveuUzY9QvDs7u+TZwWdfomahsloSHQVS6ur5J0uWENV79IpJijKUlHhJ09w/omc0696K+XyWdDDtWRRQwu6yZ1N+6zpwQ4Sh2viMkQysqKk1x2gGghNzWw6ZjoapdUDkyplarfLVaw1dMsOk8kEE9O2OxbBOEf0kbppt0YxSl641A/q8D5tqCvRkNsO3vvWek084Dz/rPGKs9IuNE67NMRgme8Ufri1+3wfsjv3zyTdscCk8//a/WnLMmcwyOl1LXVtkUIwJoCJTCaec0/v0B8HticVavrYoktvUOCMp2MN/cxSj3cYb1/BaYNF6Vs4fc9J7n34Ydbvuh/rlsnI2XzmCd534zFWBhXlconFINWMODVEX2JthywraSpLp+jTNIqYZKI756jrKb4RynIAmqEkfqGuK6rGQyZpp0eb9uxuMPTXNjh56jj9XpfdzS2uXb7McDKit79P42uuXb2CVWW4u8Wf/PF76PVAzJjV5Q6VeqaTSMgMS0sl6yvr7XIET9PsUWpGb5CzNa3YH3sGRUaeOfA1ee6AABoPyO6UvBpRDFluiHgym3gkj8cVReJcxOJDjfdQB8U3Bg2x3Z4j5YepesRAXTeJhG3XJ3of6fe7NLFJiYQmrfJXVYy1+BgpCsAFRCyTumFn9xrHThxnWu2zv3sdVUuZl9x95h5iqBmPdtnvjKknOaqKRxCbsb+3hzFTTp9epztYool75MV5Lj372+xOSl71mjeytn4fGgdgI7VUtG+PjW7+Nx8O9mdKC3L1YGvnFM0UTJGRd1boL/VZj8eY7I+ZjMdMJxNGw30EpZlVSIgQQ1oMHhXva0JT4UxEHFib432TOK+Q+K35NsWBSOWbFLiRlE8Wpd1a2QhqIBBAQ9ojy+YYZw9XQnwGx/RLkTsYmOLBrJX4pYxB3+JcoAmW6Wzc8iDQ1MK1y1PcYIMTx08gWYel9VVW1wccW+uxvtSFySZPvP8KY2Ysd5SlXkFR5iyvB5bzqyy5HtLJ2N3dxbDPqZM5J/tCUYIzGfVIGG4FmqrBFB1CNECJ95bgPVmWEXXarhA3GCnIsx4xKC7LiLFOURyX/nJJYRz7N3b4yAcfY3865HWPvJ4id0xGY/Z3J0yrQF4atsdT9vf32N28RuYsoaoZD/fpZCVrqxmhGiPqKHJHXQcmkxlXrlxGbKDbcxQDi80CZ+87zV0nLZPtCXvXt/DBpAiPUYosLYR2mcU3aZeGGHyajWtPXki713Qisa2NeE2LnL2P+CDMmkg9iZRF1v59plQ2NqVoFHlJjEJV+XZTNsdkOsUVDptlaLvAtixLJrMGV3SJJgUpxAq9fsmNrRFb2yM2ji1jJZG89Uy4emmTu+5aRbWh8RPqUHDi+AnyYp1Lly9QlDkrvR5+usOk8ljXYbh3nt7qvayYdX7vXe/m1N0P8/o3fBn50glivopKj/RHidvtcNohnbb1jRz4Q6R8L5F5qCYizmLFsbyaszQYcOP6dUb7e/imRjRihURaB5Ll4z2GgMtSLrYPKdfL2Yy6UmIQrHXUocJrwGSO2KR91NP+4QZjHC53pO1Q0qZ3Ytr6msW/N3Mod6obB3c0MCXCcP4nmTKn3HW8T5HBpEoDpZk12NDBuZKojl5nnXe+45301gYUgw55bslNoGNqLj5+mfX+jAeO9VnKPauDLuIE3AwbnqW+us1Yn2ZrJOxtbZEVNcXAMcgrMmPRvMSpZecGaLQpEqKWqgk449qNx5RqVqXMaMnx/jDUHmNKwoveE5uGrOggqmydu4ztWJ5836NoFdjd2afSyFu/8ks5duoENYHheMhk5zKZ8eRFUsLlgaHM9cCtDF2bkiGd0GhDpwBXRsAymszo54GVjRM4HXL13DUmuzUrazk2i9S1J3cGHxqcyalacjpqG0IHjHMpcmY8ISTSW7RdKQ+AofFKmYHLTLIISMmxYtKmc75u2hyfhT8QEKGe1umPMRhLDBaTleS9dUbjbTQaRHPyvGR1rcve3gg0Y21lg41jpxkPpzz+2IfZ2tyiv2QwJqfXHbC1ucfy6jL33fcQ5859AF9PGPQy+s5TzTaZTSpGw2v8/+y9Sa9laXae93ztbk5zm2gzIpvqWEUWRZGiATcaGJRgyB7aA/8CDwToH/i32FMDBmiN5AEngmkZkiXTYJFFFqvNNvrmdqfZzdd6sPaNLNGemAboSIB7kBkRiBtx45yz117fWu/7vOuzb/H7n5zxJ3/+P/Lq+R/znd/6L/jB7/7XKH8KykMJciyqt7SCpUi987CBWs5I7whdVYbQYZh49eIFl2/fkkLA6oVeWYt0ljlDyuhaljonvkRzOwSPIsY02gohwXkqhRozxlpKLItubFkwqFs7S6YxMgdTWJnpm0Uy8I6U+XeF6W90/Xqcj6SDwJ3zjpMTx/XBULIlhIqKGms8ScPbi2tWqxXbs45AJNWMSYlQ9xhmVk3ldKXZOoVXo6AsdKUxGp2umOYdKwzJHMjthLU9xsiMI6PYH2dy7cjldm3OO66PGDmPxCSivMZ1lJIALwpqBaWIn8samROZrDjperCV64srOtVw6jtoLV/+7BeMw5GsCl8++Yo8Jh596wH92rHfX3A8XNGYHm8NECTM0li0b+g2FtcmXKsIQbE/jlzuX/BcXROuJ46HAZMrx71Ge4OpRrAvCmJKxJgpFHxrqLqSiySs1CpEK2vNsj5POCe6oxgy0dfF3KqoReGMQVlDyJlpmpjnBNXQ+IZ5ntGLcl8VtYgvYZozqu/56JO/j33zjOdPPqXWQrtac+e0Q5kL9vsDx+FA202CD95usBZSGnHWkUPmWx99RMiOly+fYJ0hBzjsRlQT6bzltHOEcGR48zOUW/H9x4VD+hI//AUvfrLiwbf/ALf5iKwdpWi0kkKt1S0H/uvUFL2EWkghliJz3O15++wlu5ubd1wn6tJdIViUFCKEKLmASrZxMl01eOeY84RbcC1xlmXCv4eMNlYEmMZIMEKp1JrxVlNLpBaNs/odfliM7rdF9P2+3t/C9Nd+plWhaRR9b7i5GZhLweiekh0FTWRinI6kHBjnzFwDOUFHoU5vef70KfMwU1pQVjEfJ6rXZFOYnafUilOZE1/Z3oE0GfI8MgeZLQyHiatdwNSGTd+RUiKmgDF2EdsF6oJWTSHibHoHyi81E2MS+wu8U6SbIqA7lQvb9Yp123PY74l15ur1jjevnlIpwkj65AP6/gTnKlodyMkwHBLNyQar8uL38yjTc++DR5zdc+x3F7x6tmO1WVONRReLsj2rorl+vWOeC02ApogfsW09w2GkabwcwxZDaamFmJPM0IwW+8gcAJkfeScEztkuN6qS4XXNVVAhGBFnInKAiszivHfEmMi5oDKMY2SuIk5MuWF18gHd1Q0vXz3j4eNuWYYkTANf/eoluWq8h8TE1cXI2emWjz76Njcvn/HZL/+KZn0KrqD0jLGKmgy7kFDW4TDUGmh8oegbSdEphunVj/BU/uLffcZv/P5/SX/vN6nIcPt2vnT76ZxnicXq2mZB+mpyiDx/9pyriwt0qnhtcE6LLcUacpaggRIyNWZUWciVy5xNAbqK7sg5B7VK8a+FmKRLc9a8w7qUIhIQra0EMGhNjLeG4mbJGbz9jv96/Of7e723hSlrDdWii0apyDgUdLHc3TbUeBRMRwkkZZljwSnHR/ceU3NmPgzswiXOaA77A3XcEcaRYZy4GTV96+nanlThOEVS0rgmYM1M5w0qGbzRJJuZDpBjy/7S4s0pGkvKQUysSbZZpRRyjGidUCphXcWairGQc5QBcQlUMv2qJ5fKmBPOOMYQaLTFTImrq1fS8hvpIBoDrvPCmabj2ZNLMoWb62taB66rVJ+o2qKKIgd48OgD+vaMYRh4fRGIassnn3yPUhOrBl5//ivMWMih4eJqJmVHKZ5pmrBOVtF5FhMzDShfZI2/8IyUltgrtPjpnPLUVNC1YpQ80U3TCOgsZ0oV7Ih2t92XmKhd4wg5UY1CGSuasyFDduwuBl5/9oa7H5yTc8M0VH7+0y9QTHQtrFYNd04dF6+ec36+ovEtqw8esz9O/PlPv+C0nenXisPhJd5blLVU3fPgo8dYZ/nsFz9h20lXkkrAWffOWBzqK14++ZdczWdc3LzlP/hP/xvuPPg9kvLoxeun0Gij8Y3YT8ZxwmjDcX/g9fMXHPd7qBWvxJNo7SJ2zJkUJjnClUylYJwR4gXye+U1y/J3KL8EK4jWygidkJQizoHzRsSpWLzzhFnoBwVLSEkKoloMMLcD/CU5+X0vUO9tYaoKKLf59Yo3r6/4/PNf8PCh5qOPPa/eZOa8JpdCKkca7Vg1hsPrJ7zeveD19Su6ruHedkXvIsZX7LpjUoVJdVitOU4TN1PC9vJ0UqYSUqFRsm71naJky/VQCcFBbWWQnY/CjS7CIlJWkWOSr2mEdllJ5DxjndAdc4SuaxeoWCEXSHPEeUPMkj02DDMlFRorOqG281CgpMIXn37JcQ5UW7hz74QH97ecrjXj4RJrPHOZOQwTr9++JLvEcd4zx8D3v/dDUoSUE0MaSGlEW/k+lUEc7cYR8sxhP2BUwShHrfLkXa09c54xTt6HnMXYm3JGa8M4BkgQx4w1UsDGMGGtfUdmfDeEXY49WhtSkbXGHCJKG5rFw+aMIdwM/OhP/i13Hm44O+8533ZM446+t3StpvWGu2eWe3d65qDZHyOP72759nd/kx//6MccxwOrVnPvzgm1wFwq2IaXL9/wg+//Bn234erqNZuV5fRkzeF4kAPUMi9adaBbw6Su+Hd//D/wn/wjw/bub1LpyNVTVabmmTBPzHNgHCeOhwPH/Z66sI9YMCkKJRvOWoUPVUWeUUqROVEuGOsW3G+mqITTWvRPS8qOtHPmnQpJ17J4DKWrNdoSU0E7R04Fqy3ohHHNgpT+v91df4t38t/sem8L09eoNw1YhmHm8uIpv/O7G/7xf3aH/+1/veLZS8M8R5wTNtOrV7/ieLHhL3/05zx5ccGjxw949A//HttVJXYdenWO15rqhQ7w7PkX3ISZR9sTzu89pG8Uzz77gmme6R20rcavOg7PIqlaam1ISczAWmVSmQWorxXGFUoRp7rgeQuldoBsscySIxZDRuuGVWclCpoEZFJOlCzHHudbFJlSwJRFOxMjXmm6zZrv/+AHWJ+xJvD04iVNdpi+47zr2B2vufrVJblW1qs1X/7s52SliYyUfMPZyuA7izpGrBN9jdIabxpiGmTjk+WoVXUhhcrmZIMylXkel9nE4vnKihQLccjkUKlFtEtVZ2FhqSVZtlTUgodVi8lUPLSKEEQdblyh61tiyty953hzdcSqzKbRbBpDXPdUIikFxmEi50LXr7j70Xd58vyGn3/2jONh5sMHZxynyts3r5nHxJ3zE5yztKueq6fP+eyXv6BvOw5oxiFi9YR1jYghU0RpRykWZxU/+OGH/OlfPOHf/Mv/jj/4J/+Mfvv3KLqj6glFwpkG1QgGpfGeVd+TUyLHRAiBMsevTbqlUI34MoWj3pKTDLtTSouUwko6il5ivoqiJNF15SI2KwWLQXcJfCVjtCNRKNXgjAKVSAWqNtQlzPObdr23hel2vSGxSobTk54f/vAed++8YXt3xOgVf/RHF0yjI6uOamc2pxbfBOZp5Pp65vzeyIu3T0kmopqB0sF+d+TqaodPipsp05ye8dF3f4f1eoUpBetnpukNOQdKgjEqhqmCajG2b07dsgAAIABJREFUgTwS5wlMwvlKKYG06FLqEqQoT0ON0mnZglRco6lVtDwiYBTaYK4RpQJxriiMzJ5CwjsNpeAqdE1L13Qcp5HtdkOcJw7jwPXhDRmDsy2PHj9ahJ2By8trbm72hMPAqzfX6MbTbS1Nn0BV5iAo4L43zEVInalK5xJDEmvNGDGdxjSe3c2McaJ6Zhl0g+A9SrbMY8Zqs6ylZRbivPjHUIZ+1ZJCZLcbUargrWSlDcMknVkQ+J61ikzEeMX9B4a+KdR0CShqiiK8VJWaC2PQTAfN6ccP+O5v/xb7+f/gs6e/4v6p4+z0hNPTLRevLnn+4g0nJytCmOgcXLx8zenJBqcbcg4ch8SqF79f0YmcKlq1XLy85qX/lN/68BH/y7/+M/7V//zf85//V/8tprNUnRYRpBh8m8a/Y4nfYnVjjBAT8zQzzzNhnkkxMs0zOUaImaJk7mbNsshfQgdSXXDAqpDKgo4zDms1pIzEzGXJK1SWXCQxBWOXdOBKVYaizLsTB+r2bvpmXO9xYRLkSUVa47OzLd/51odsNkdinbl/r2O1Oud/+sMrnj3XYDXKQraVR985Q50oNmcdl+E116+uObkjrKHpOOODpc8drj3HNic8eZ74q6tnmBTJwxE1j9zZWPRmxW5fmEZDoy3WSHtujCaGGd+JTGCapndc6IqQC3IW+UCp8iHWuhJikLmM7jHWYrzleByZw0RK4gIX8L1aBH4ySI6T8LqpCV0SL7/6ikM8gst0q5ab6x3rfk3fd5RcWDdrog3UOuNXjnsP75P1RGIvHKGSUVpjXMbmIvHVtWAMpCBrfuccKRamMZN1ocGQnNht1HIcUxVqqpQE+TYGS85tjMOMcw6tNOMYBIeMbOtSLjR+GfBaQxGCHIWCNoW2dZRa8KbKLGuJWY8hLsceS84tT7+8Iepn/OB37vJbP/whn+orTL2i1pnGK87OOkoqdN4yzyOmZE5XjjzPQnY0ljCPHMuMPWmXjbph2E/0puOzv/wpH38Sebz2xPkpX/74X/Dgu/8x3Z0PwfUiEVkYSu/i6o0A97xt0S10m7XYRaJs3EopopCfRuZxYhpHwjQTQyAuqbm1mHcBEHnZCBrXUFIWfrqukpOnEJFuFsuL6J+WW1oXclFoJZYrIWT+bd/Ef/PrvS1MqniKkmFyWYIC53HmfKNYtxb8wD/+g0fAij/85095+irx9u01VbfcfXhC7o70Z5bjfORmnEk7hVPgqqXTDlM6Uux4+sXMn//4c968uKAhct5n7p8U5pvIhcrk2uLqGl0UlZmmAV0sQ5CWmeVmyiFTi0FVcaOmFEmpAhHdqGVgXsAUcp1pbL9QDZPEiBcIIaC1wSuDbyzeKbQCbxRTSVgDl69eojzoxcNVjgI1++znX9K4ZkmBibSdQikJiozpgHZi2FWqkpLIV72H1rV459gPM9MUsEbkGdMUUd5gilnW+oVpSlgrHdw0zdSiiKHKv1vbRY3MgidYZkhJrBi5KLnhsnjHYgqUWrCqLp3eTOM81hhyCjhvqBqmJMk2WhnmACYJPvdkteKwh8//8sdcvXjJJ9++h1WVtuso44F5DBhl0Ua8ljEmSgx43zGkTLc+4ZPv/AY//cmfM443GDVxftLRtg2N1hwOkbON4fkXP8X5Dfcerrl+8kf87Cf/in/4T/4pp49/j2yFFMDtHEfxLmeuImzukjMKQ10Sda0zuEZRN1tZ+udCjol5mkQlfjwwDDeUkpmnI0bf2k60kCedWmLPtVhSYhWUTylYLdlzeUlvuaW/3sakf5Ou97YwVSzUBFqMn8fjyJMvX6DTNXc2is09B+6S//A/esBh/pA//OdfcnMx8/OffkF3NjPFK+p0YIiy0k3R4F2DqtC3J8QLxe668vz5wOcvbxj2I9u2cvaJo+9bWt9yvAz07QarWqF6IU7vGGdykQ7IGCkeqRRqURgl2xHrG5qmo5RRhG66ElOk8Q5jYQ5Hxikwh7DQDxK1ipZl3bS0nadrDCoHiagqCqUrjdZUFgolShYE3lO0Y38z0/drtI/UfMR3iqwSUxhojBh0Q0hMU6ZGaBdcsMy3JGjRGCtcqphkSJ9lIGy9peTEnBK1jiKcVJY4S1GNUQbdpdzOBsGuNI1rSVm2TEYbYhgJNQIF5wyVhNIaa+wSBFHeURBqtQxBjseNNSjjBTBXDE6NPDpTtMXw5vVT3nIDPnH3Ow8peUJlRc1y3BwPB0gFr2V7lXPl9asL7t5/zPn5XV6+ODKPkeQVVU3kUlAq4j08ftxzfZl4+eUvePDxlnAd+ezH/5rfPf8+oRUig1YiIBVq5tear6oBrWV+pIXJepuBmanCZ9IK3WhWTcP65JRaEnkeGMcDh+NOsCghMY5C+aw5M48DtjPEKaGcJsdK0zjiHCVKynjpPrXBKvtrqSvfnOv9LUz2iMoGWzoMkWmIPH0KeepJJwfS2OO2I6y+4Pf//kO+/PQef/xv3vKLz6/43vYOV1Nmuy50a4PFCaHQtXB0XF8q3j4LXLwZeXs5wdDjc+T+WeHb31pxb5vpqPjaMO8sjVM4E6h5YBoO5JpAZcIY6TqHxgjUKxlqUZL4ofVCO/TC3qkGjcHZFbXCPByZhxGlDCYrcjTkmuhWLd47AdpbBcYy1wxOyfxpySarWaEotM6BUgRVuffgHo8//oRUJ37+iz8TDVQx1CGRq7DRSzUMA9gIHo1ft6hOY1aKGCeoFascoWTJ3htG2s5hkjyxjW4wxTKMB6wu6FSoUVHmQs2yjLBGyYzJZHQPVSWU0oQ5ojVoA7Vm2taRiyEnTesdxsqRd5oGUALgX7eWKUVCmfC9Q0VFMYGSRlxrufvA0q4MMJKz5c3n1zx8dMoYd5Q5UvOMQqFMT4hrPrj7fezuLW/efMbnf/UpDx/ewxZHiJGr3ch6074LAoi50NvK6rRQquXpr/Y0Jxu+ev0jPrr5R/jybYpp0Fa6PuckHFMX6WhUkllcrXnpqqTz11pjUIsYUz4nRYk4VVmFtytM29KfnL+Lg5rnmXEa2O9vmI4d43GgmJmaCrjEHEeqLwsAsaJpRcSrNUlD0kBdglvfzW/f3+v9LUx1gcQhptDDVPjFZ1dcXAQO9zwPB8v2fGb7MIOb+Ae/9zF/+leXDOMlx0mz2jo2J47GS1hiGDP7AfYvZ/avZq5fZS7eThzHhEmKu+eGs03LdtNydq7ZuJ5rNTPuZBCZ00RJMykGqgbvZdDofUMIUboaJVleIUSaRsB0KUVQEt/cdh3OeXY3O4bjcSleywwF+ahsTnqUEeFlQVjQ3ovmxhixKdxG+tS8cLm1JZfKgzt3iSHy4tVz0lxkgF6Xr1lyzlzToNREiIGueCriTzuGhFWWHCIxJeYpLTHelTgLX6hbN1Q0KVZKUmRdscbJPKRU5jFRiwJliKHQbxpSiBSlMFaoBbFmJJxU4pK00ktWWyVNAd9IbFLJLFFSTrLkWglMUCXTesuYNMZ6vG/emYQphrdvrvn00xusMvTeU7JgZXMyXF6MnJ/OeL9i3a+4eHstczAtycchJoYhsN50pDQtpmaLQlbwnzz+kF+9fkp7XpmnG0J5i262WC9WkWAU1nu8a9HGvlOD355ub4kZv67ezovsoi7FQimRHAnh0iyBGgXXNGxPNty7c07JmeF4JMwz+5sd0zgxHg/EpfuOUYb4NRZKmBZD9tfwxXf32N/mDf3/8npvC5PgPwtVzwJmdxv2oeXZr97y008DH97N/Ob3Ko+GTHd6pJYjp2dwUWaG8Zp7H6y5c39N33pI8OlPnvPss0uObzxXzxVhsIxDoSi41xfOTxzeZsI8UGtHt2m5vBjQNhFLwlFoGkcujjHM6FtoVxVnd8qZpvGMJeK8RDcbq4iTYGi1lvio4Tiy3+9JKdO2dllRK4xStM4Q4khjHHNMxFywVpEKmBzkQ7tog4zxVCXspDwnSrXs90f2+9e8ef2S7UY6MZXliJljwSqPWYajKVWmkLFTQM1LftoyKwlzYL1qyVGQsbdUy+kYsc7inMXqZjn+2cU3ljFVi6l6FpyJMx0xCAMqTjMs8VlzLFAdJSnswhpXqjDH2+RjBctcxnlPv24Zpj05Czt7mgZK1uyGSN+ZJdFENorrrSFmhzO9RCoVi7WaYSwcDom3ry9ZbxwpVrxvuHh7Sb+SOVpKhWEI1KpwriOnmf1uIEcYdhoTbzAlsL94ThouULXHFAWpwzhLLjCHwGgmXNPQNS1am3e4FKVEnHorm6h1CaMEQB42pYgPTwr3Yh+5xa0UMVhba2nOzkAp7t6/L17JEDkejwzDwG53YBompsNInIIYeG+xut+Qvdz7W5jQiykyUTTMWXN5sNwceo5z4mI3karimOD8QeVyuCaETFGK3T7QHR0PXcsHjx9Qxom3X+55Mo2kyTAPcNxHrFecnngensLd84ZVnzjs90x3IJWZWCeqzuSyZHBFeaI5URJKt1LVuw9RTPO77VHTeEKY8I1QDL33xJg4HI4yRHaalKRNd85ijAMPqQRKQVIzKCSjMKsO2QPfprZW5hhxXhFzQVvFzW7k8vILUkz0rcYpw8lqzTHuyUnSO8KUaZaIpZwrKRWOxxHdabQrYhi1juoqNWVJTVFyrMmlEkMWpGsqeC/EzpBv94d1sUqIhisVqNWQAnhjKEG6g5SrpNBEzZwjp2crzk7XzGFknCZCKHjXMwyRpu84O7lPUhM3u2swCqe1DOunxPXVQIyJ8/OeXPISvpmxygPQrXpyLjhrGMfM4TDy9s1bKmtqkVCEmKwwy8uEbA01MQjMTStJPkkhE6fMzfiW1Ykiq4lh/5KT7g66npDHINYSI7HjOWfi8SjWJOfwvpEj3iK8VAuZ4FaIeUsngFt4m/RZ8tAz7zZ+tWSMFu9bLgW0omiFtg7tLGernm0p3E2FEgtpChz3R7YnJ/J3/T+KLd/P670tTLUaFFnwrAqGUfQ+w6w5BikUo15xPRvybstXzwb2e8XcWi4uMnt9QzEVWwwnjUFNlSZrbsZIGBNGaU42nr4vbE8d660llUnaaAPVZnRTqVqyueKcJa21JEoOqGUNK+D5SkmJWmR1q7XCWiXBmfo2Y0wiyadxBsT/JERITdu2GG1JOlFjJMVZjrFaif8qZmpWohui4BpDLXmhRRq8s2zWMAwzq77BW03jYB4H2ZQVjTKSPGy0xRpZ9WsMRlv2N3t8ZzjZdKQQ6b1nKjM1J6yWoXOmEHKl851IC7ST74kiBEVbUKmibBXEq6ocp1Ey6aIijkUoAwVAkVOlaXuoinmOTHMQGUAtUBvCVGlsQ9hVokqUANVDTJGKsNw3W+EUxRhA59tPDjkF7t9/yIMHD3n67JeUONK2lcbD9dUVq1WiaS0fPHrMZ59/IV1MltqvlJAg97tBCAtFKJ3WKBojCBl74rl8+wS/fsiqv4uhEdKFF0xKUYVCZQgD3ntSTHjv6dp2kZPcRoIvSwJrl58veYhKk3JZOiX9awwoCc+s8C4nrsqASr7v5Wu1EwGmdQ1N27Ner5ff900pS+9zYQJqXd7AXDnsB9F6lCWIsdcccVylludPjjx/uucYoHaWeS7s31acG5kuvuBe35LeBuIuk46VRivW2w7nM6ZGCpUpi9CtakXIifVJx92Y2D2baLSTFWwRqqNrPEUVYkqkhSSolCSz5AR+2bwZo985z28d4CEknDdLtI+iaQxNIwbMkgP+Viu1GDqtMqiiiLPmeEy0vcVYhWksykacU9Q0se5aYhixtiDmWkMumbIUtaoUJdZ3UPySIYeCLppcK3GYOZZEYzTJZLSqOC/oFmc0YRZ/oNEaaxzWOdarLaUUbi4vZTvZgPIWqzVzSsx5RjvNOE6QNMbZxUEPVVeG4cidu6ccjztyiUtqbCHOit114fLFJePNhOoCZmXxxkDJpFCoFE5OekKIGKuJsVCrEcHmlFHVcH11QwwRZ2CzbvjwcSbPma6HaR4wptCvWqYwSJCCkkAB59xiH1mOUabSbzydcgQ9ox24zlAJIhEoCqoc24lF7iq1iCWLhAzkOTAdB9q2pWkajPdi/FV26ZjUu1mTZBzWd0e4dz+2y7F5YVep5ddBZAqwBPxVLds/VUlF/HMoZGv4/8O9/De53tvCdDscXMbfGNRinRAeeK0zLy927ONImBPXu4TpLatNg9703Mw7Lt5k6g4mlbjv1zw8bTlzlt1Jpml75rjnMAZikra47Rp8Ix3PdrvC6sqX7jWN88S4dARKYRtDKhLlY4ylInoemQEVttsTqgqUkpYnpFvW9NJplVKWgascn2ToWWmbZeZUFy5zqSIjSIUwQZhZFORfx0nnWtFUnAXvEL2UVQsNQI4CoEi14I1jGiaM1niv6ZrmXawQyLA0lSJdWsqkUvDOk1Kg71pijoQpEGLCtQ1tW2i84/R0y/XFJbeJNhVRLzfWMk+BHDMmKUqJaGOoQNd65gwhBoZxxjpP17XEPFOzYb1dsZ+PDIfItm/l2DgnGmsk9cNljAOdJQvvdm7jnWcXZ148fY1vPSlP1DJx9+Qu222PM6IpOwyR6+tLak2EMGKsvOZzSISQsFaivFYrTwgSGV9I4l/TmhTkvUTp5SgrR7JSC6YqqhJNV46SFycGaE0YZ4w1+L5jtT7BmuXIphZJwS3G9691Vf9eAVoKl9ZLTJSoRtBoqtLv1N/GWqpZOi++Zn2z/P99Vje9t4VJ1wIqCT9QyQamOoN1dXmRLdNNIg2RYhTJaKIqrJyi366w1fP8q5f0ERqdGafAmVlx1ll++94djnPg2ZWiFofXgmg9OdWcnDo2W8U8XXOyMfRrSONArRWrQZsCVY6YTlWUyjjfoKoA9n3r8I2XFX6SY5sxnnmOIqQky7oc8UCh8te8nBCly8kFYwzeO5SClGSArA0oZXCNpuiAcQaUZc5JUlqtGFGhiHmzKIwcTvHOQCrYWLCdxa0UIQaaLE9hbRXaig2iJo0q+usUY6WoMdA5R0hgVYWQGOZrStdycrKhX7UcjwdKkRv0/M6KWhXTYNC0FGZ0LnizkDLnmUTlMAWSW7PafotH3/6EFy9/wrPnv0KfNpysz9is76N9w356y5ze0rWGXisO05FxmoWlrbR016VAjbSbluvLa87UCev+nOfPX5HnI75RFCs0hgwcxon1uqUwyLwvy5wMLM45ak3C145JAieV5ngQAeX2tAN7SqDFOEcuoqBXS6CD0W7RYxWUqqLqNqIZU0pRDjsOuyOr1Qmb9QnKmuUsKUufpQmjgnweFv2aqOvVknyycNiVRiDNFWohaUWEJTNP4st0rehFQ1VEm/l3helvcimWd2aJbrJGvF2oGaMEk6GWdI5coCpLDFnU313FdhajFONQ2bY9r94e2OUDZ10DpbA5PYVdJZOYciarjm6zpekmfAP740gO0Pea/XFEaSMbo5LQVlNA8uOcRSmN9QqT0pL2IZjUvBwFrJVtHCrhvKLrRLNTySKuzJJqm6KIE2/Z0rfYVucMacnIs1bRtJ5UZ0lX/bW03JTkNZEY7iWKKdfF0S9FRxtwXrFaNxwuRsSdvuxqqhwdY5BkWawMdAtgrCZMYSFKFvnzqKQwc3UlKbtqOeZtNit8a8mxsL+JHMcRv3yPORfKHDCdly64wngYidMblNK0qw5VPXFWnJyc8uF3v8fq7JTD8TWf//L/lBgqY3G2ZT4eUFhylkVESgmtNE2j2W493mmMUzS9pVm1dKuGabzBGEdSljeXO6yTGWbft4RZZB21yGt4+1C4hf4PKXIYCjfjNR/8g/usNveZI6QaKapK4OptwdCVqjzaWeAWFChdlKJiM8z7A+EwM+6ObLcbfCtD8rIIdllkFApItaK0dG21Vik278y5WSQIpcq2zxr57LEk8Oqvd3F/d5T7/3jJ26FRiDitcbDqAJVYrTzjsZCToipHyIqQMqvtmmyPxPlIKBWjKnEqHOZKvlEcpsLgJw7H19wNlcvdiLYdvtFMc+J4mKhpgBAJO3Aq4pSmdYo0C9aDBRRmtcE0wifSxpHmAW0qXe+JaaBUkQuUEolxpNQZZRJd59hs++XXZ4yRyG2jHcbdmjPlGGWURWlAV1iKirGCZrVGEVOk7z0xZ7w16K7jcDi+WznnUnDGAUWiqoykmhhbaXtDHCSMUl7tijdOhH86kxZR4C0+V4pyxdZKilkG0moZgheZzVWUhA4MIwXPPAWZkSypyClmqqpYbelsQ5xnht1IzZonT7/ieDhyet5CaJh2E2cnjnGY2B+f8ubiK477IyeNZTxmETOqjnGaFq+YpIGkUnEmCUAuj4yhUnTBrzrufvCQm6uW/c2ebuXpuyLCxBxJKRKjRJjnlNhs1lQy8xwEYTsv1MlqGQP4dkvXnZFUQ8qFXMWbJlYQRVqCM1XVC1LXyeusCqrIW7oEwDDv97wdBpqupV/1NH0nivECuSzdUimgJdYeIJb07rgPy/xoQfqmFClKY7OCIZBilPfyG1OW3uvCJP/VVaFVxTtF6xP3HmxoOsPrlwPHYyUXSwqRnCutd6zv3UO1E6kGtI2EUggho2JDDokUExmIbycud5Fu29N2mjxX3r4c6NuJ0GXYOs63WyyFlPZ477E5k9NMWuLG61KUcpZNibEK4wXkX2pF10SlUGvEWAnIbDuFc3mxPdwONyUKusqqDFXlSRmiDHaFcmjwjaNtPIosXHGALFuw8TDgm0aQrEUsJKZKaTfWgNIoI8dI11iaRqKmrMiwqVnhjH036+iaFlhuxpSx1lFUQS/GXd9aqnypbNOS3ISUwjRMxBSwxuKsxRoLIcvNtnSB8zgzDTOmES3YutNcX1xw8aZwdrpi2M08//wJRJjDyGF4gyoDu5yxxWK84cOPHzMeXzGNk4RlWkmqMVUy37RxpBTQRqin9x967j94zHD4FOsCRinmIaBKIcVKiHWZ90iMk9b6VqUhZtkEKEXbOA7X13ynX9GvTgmpMswz++NRAlaLJKzcJrmVUrFWvVNl15rQylLJlOV1qrUQjgN5DgzDQLde0a1WwvROCWesQOCqJKNwe/xfOFclJTnlocmqysEuFtSiHK9I9/5N2cu9t4UJLVnwouhQWFU43XTcv9OhbGZ3rTgeJQZonBJzrLx+eQ3tlvOHa8JwYPdqxE4KVytOO/ymI9eBWWecgmwcl9dHUnLs99C2mfWqMJ84pkGx2xWsS5y09+n7DeP1G5QzEOMSq+NofMscowxCF7uFYFYVuUQArDPLcUuzXntQCZ0L1kmum3Nm8fQBBZyzxCg6I+fc8md4hD6oqVWelkop5nESk2atTHnCOIt1BqUr1hh5oqoindFtaL0C7RR+baQIBikWKWYBkwFRmMKkmBcrjNhK+rajqkRBhtC5Zpz25DgJlUAbYoE8R/pNK6RKl6lZOj1lDEppUshsVyvGGNGlsmo0bWNAOay2RDMzXl7wxfU1TefoVgqjCroa5mPBNjDsIk5vUEun9uDePVKeCNMlvnGMx0jJEEfF5fWexr7i/KxZOpXCNA7klDG/5mHTS2BAraJh0oulI2dpcRoDc5x48rM/w+p7nN37bc7vPeZO17NebzmMgTmIv7OoAEhYhHSeRlb/WhELLOcswfYiHQ85M42TBGmOE92qxy1AuVsvbs23OF7xJqpFpCkJPYlCBS06QEOSY7aSY/M3hTDw/hYmld+1nwponOPenVPifAVRYVVDDBNhNqToyCkTdoUnv7rmcBMlrWSynDU9OczkfKTtFabNZFWJ9oBbK+abzNPXkbazrLeGM9UQjGEXFQdaHn5yh/vf+z5dzrBeMe5eMVxdUlVls9hlbtk4bdvJkNxaWD7UNS/G1prp+w7jBC9SihwbjLHkpEhpsWssRzfXOPIUpNvTBmM0wzDhnKPrm8UyIagUYyw5ZlJN2Cq8HtkCFUBmJdqIqXScZ2xwsGiPUsi/huswC2upoLWkxGploMqNr5XiMAaqdqAt45QoJYqFpCCr/CxdnzGaME7EcSbHDBmJFDcSfNB1HfM4E2PCth7fO6rSpKQJU6RvDSsroQ3aSZBnKoWKJ2SIY2IaEvOUKcmy2W755Nvf5/PPf05MlVwTcS50vsOvOo43O778xRPmhw3bk44SwqI9A7Rb/s1SnWRMVHBO8umM08w6ilUoR7bGw/CKH/3xv6DGf8vZ2Qd8/L0fcPrwMf3JPe6uTykYkvfEnIhF5BoYQ0gJVCGqBKVKQkpVqLwcyZbNZk6Jw+HAcTjStR3r1QpnlWwGF6lJVbfGYHnoLPSYJSa8UGLEliKePLUUpW/Iae69LUx1ecCDDHgBhnEmxCO2aXF+C0oxh4pzLaVGVK2UqXL1IlConGxWYDV9b/nudx7g28zmzNFtDafna1T13FwFnj133HvwIdvTFauNw/nMeu05Oelp1p5N19NSWH34kN3rLxi/+CWHiwuGYY/B4JzCJYN3nrIMr9FQs0QitW2DtmVhgId3Hw6lwHlHtRI7NMaZXCveO9luVchJBppKi9guJ5n1pFtsrZKirLTCW+EYKeTGan1LjBOu8WQyMRWcNgzHmZhntCkykC5VjMZJWExlYXSTiqzFq6L1DSlJ1p+1hnGcsM5Sq2YYwtdzEKXkSW00JRQoljAndFQYq0ih4LwlRVHfF8AqAdgprclZ01rP9f6It4b1umMMI/McqUq8dcNCh3xwX2Bzby53PG5XPHvygpv9iC3Sfaz7npPNOTW1zINitzvgHJQcqGWJrAKMNoKoWXRBIuEQgWwphbqIHEPMwlM1ifv31zz5/AlPf/FLTjf3+fzHP8auTlmd3OX+o4/48ONvsf34EZvzM9qmoRpLqAXjPHOKQi8t0tGqIsNslo5GnASyCCk5c9jvmIYDq75hte6oSlOLXo79WY72pUpxq8ISz7Viigztc863wsBvjPT7vS1Mt72SQqKi55B49eZAptBusihqlSfbibazuOUGmKZAtZqhodrQAAAgAElEQVQ7D0/51m/cZb2eeXDP8ju/+S16r1ifKDBXaDtzOBz4kC2f7O9ydv4hTXdKVQ2RTLO2uFZjkiXblkFr+vPHfPD4N+juf5tf/Nn/jj0+YT5MmFTQtmK9IUa5uXMWxbhkPmesrqQ8SmJKqkuhUbhGMB+Ny5A0cyw0ncWYBj9lDtdibSjJ4IwVvIgCZYp0J7ViaxJKAU6QtVrhlKVkibuuWgIC0pAoqWCCHF9crzHGkY0UP40izLMcQ1PCGrMcJW/t1AlDhRLxzjAn6ZZSSBSj0MqgjdxM1jiU8qQlTVYlyBV0UaSiSFFQNFUr6lSJY6Lve1QtHKYdpoq52HhLPmbmKWG0J5LJKRHHwouXNwQUFMuzz18yjUeyS3RBJCVDkMj1FI8M88DmrCXWgRqFMOqMgRrF52eAxb9WiiLnICQEDZqCt57kNCFESgxolVi3lsPbG44vRvr1KXfuBcxu5vnrN7z6yx/Ddku32XD/w0fcffSQzd07bO6cs2o9k4aiLDFroYYWhW+kUFlt0LUwpbSMCAopzez3IzENWNfj23bRMc0ogjwEY4ISJEPOWHI0kAw1S/STqjJS4BYe9R5f73FhupWALTdF1RjbYYziMIzUYY/2ms40NOsO53sSENXM5u6aex+s+eQ7dznZKk7Whv5E0/uCazTjaInzkcM0okrBr07QzUQhY12HVg6nLX3T0PQdXb/Btx3eKsp4gzcbrq8nrp8FupVCD4nLN68YUqB1GZ0TiHGCXMApi7GOmCbpgpZVtHWamISIqZTCekPRUGrEu4ZWO3J07EIQhnS2kKus8isYpeRGqgaNYw4yAL5V3cUkYQdKa2IO+MbQNZ4cJrSu5FgwXnjSeYroJNZ2GbBCiqKbmYNkl+VSwBpOTrYEcawKygSFyrIZkjW7IQdhWaekUVXLnGTZ9E1TpNZISIVSRQKhVGUfDqLdKpq2kSip/eEgZuEqg3jRtFmyrwwhop3DO4NKmXTYU2xmTjKQ3x8Cr9++xThPUZZjSNQ0s91YVp2jUIW7rUWvZZQScWqMywORhSxaMUYEjVorcirs9zvu3Dml4xV51gzDNWkXcc1rzu+e06162uHAfGn58ulnfKorpvP4Vcvp3Tvc++S73Hv8mP7sjK7vwTlSldfcZOmWvbPEWilVU61FlYUekAZCiHRtS2MVqip0DtQSJfVCKdFLJUONcgd90673tjB97bleZAPG0W/OyAyMg6Fiab3l8Qd3OX1wD9tsqN5j14rNmWN7Zjg5cawaS2stq0bjjaVEhfUf4poZb0dqVvSnW/r+Pkbdp20e4tsNrlUYX8kJxjlx3EfCPPPm2VP2b18QR09xH9B0a0yTOV89Qtdrwv4pcZjxbcM0TvjG4L2Vmz8VAcRXu4RhCog/LpRH7TXaSpNVVURbS9NruqiZriILrRetJYI6pShGZ+0ZpkLKBd9aWitFRH7ekEvm5GRNLDMpTlivqVlU3RpNTUBVlAQOK4GVqgBahufLE1YBTd8xhkCImWkIUBalc8oUDdZZ5mmi1Ip1LfMxUoNiTkFSZq1BVQkksMoKnXSZc8WSqUGOijEnIplqKzWLx09pi4zAKg+//R3OHzzgq6++JO5vWHmPM4lEomaxqDTOo6xjrpa310c2my1rc8bN1TVkRbO8v0oLfTQXhSkKYzzb7WK3ublBWwhBot9lRV/Y767RtHhlCFlD0ZQxMo6BF4c91hlO1p71ds3mbIPzhjJWxqvM4clnfPHjvwDnaM/O2dy7x9mjxzz8+GPO796n60/orCcUxZAyRSsKor80SkzlIQRiCDRW0zqLsrd3jfg1UylY6W+/kdd7W5iot2PvxeelIAPHEFDOY7Xi8ccP+PYPPqY/O0f7UwKOfus5u9OizYTRGacMNVaGAUY8lhXeNWw3DdsPOvquE4+XatDqlDBb5jCzu9mzO75lP46kXMlVFLhOa+48+hhvPuJs/XtsbU85Tly8/oqbt58yh8wwVGoY6fuexhqKKuSSiWX5cGtNWf59sQgcjVpEKKhkThBzxBmHtoqT8xYVZ/bXEZSlX/XshmvSQpfURjHFQphlOxmzockV5+1iTs3Mw0DTW1Iti3+rkmqh1iT417pEUQcF2iwfbIXSBhZmlHWeMCWq0UxDYBoDFiloRskmUGuNypYcMuE4M89ZUCtFUxEpYMnSKZWilgQVUEaOn0ZZdNWQQGtLivK95VglBtt6TLWsz+/y4JNPKM7wsz/9E2wpUux1waBRCFWg7zs29x9hX7zlo0++yx3b8fOf/ohajqQYKUUsOSFWjocRpeD+gy3H44H9bqBWsN4QguS61cWAAomcRlxjmVTldvKsUOScyCVyEYb/i703abIku+78fucO7v6GGHKqzJoBsDCSBCc1SLDZJEF2t6QmadatXkhLmb6AFvoEWmmhb6GdNtqwTSYzUaJ1owkSJEGMBFCFAlAFVFXOkRkRb3D3Oxwtzn2RCbVW1CZhRt9URQ4RL99zP/ec//kPPHryhP5hz/r0iMXRgrjo8N7T+4TOmbS7y/0P7nH3W9/jncWCuF6zunObV159k9uvvsH12y8T1yuqc8zVM7fAzGYWwHaamWaH74zqoVlQ75hyIYrAaPgTyJWs5efhemELk+hhjGvUfAfVK9UZlrNcBv7Z7/4Kd16/yQ8/uEuRxHpYsVgMDH6Bk44+BPo44BeeftmxWh8RZMXQD3hfTFqC42Izc3GxZZour8YX7xXfOU6unTAslogLeOcJIuZTrUp0HVKgWwmv3HmNa5ef4J2vH7H1P2RzeQ87vye6uqfWiaTgRO0ccyYBQVqgJEbIqwIu2MiQNdF583JanVr+3LgvXO43doI6G2+6WFgd94RxwfmTHftx4ug4cOulY0oekSCQCmmXr3LqvEScBqZ5RFQIYtSFOgFiBSSnbEx0B94LqRQylnwiVeilIzdrXQWKZOZstigUR5kLUryB35hcJ00GeJcCuRpTW5ArGY6RMZsveLOn7Xyg5kQp4KNp5p7cf8Ly6DFD6MzxYHthnYLaijx4h6NQ80hJO269dMo0b7n/2D7j4GGcEyHCfjfjfc/tl65TsaDIeZ7pesc0FXIqDP3ANBnuRihomal5ph8CT3ULYsZ3SBv3qjJXi7lPu5mnmwe46FkdrVitlpwerRliT/RC5xx5zrgyMm12fHj/Hj/9xrdZrI45unaDG6/c4fVf+AQ3X3mJ49M1NXhmhO00k72Fbs6TogUCjjHPqLctr5uNh/Zzg3q368UtTAfCjVHyGNPIxfYJYaHcvH2DV1++xnqtPLz/LnlM3Lj9Cv36mONbp5wen9LHgcWwYOjNc1o1kVNlu5k4e/KUad6QcyHGJeAJ0XF8bSBGoYsdIhEvgekQwdRAySiCaytdNCDBUbxSQoD+VT7/pX9DHUem6YJy+Yi//j//NzaP3oMZOulwZFwwmUpVi/bWpvymbbMQDHtyShHFO2E4chzrQDmb2YwbuoUSuogmRUIhSE/wK/Y74eGDp3TRsb3M9IuR9VGP5gxgZEfxlKKMeTL5ShvZSi7kSUE8i9WSio2ZWjEXSudMm6iFevChysZqrqngqBTJlFTJWahFDHQt4IO3Xy/m32inuHkLmVGcAc8510ZCtNdRq5JDMcti50j7gsfz0Tvvs3tyyen1I9y+wKhIcBYAgJCy2RDvNzv8cMmWwMX2IZsPLzg5dtQwGc5UC0Uqn/jEmzw9PyflRC4jpSb6ISLiiN3Afm9ayL7zeOeZ94lSZqbxkj5aYS1q96yFThrQXEqjA+Ao+8LF7ik7v+HMn7FaL+n7nsVqSVj2iGZiHzmpzcJls2G/2fDjD97jR1//a4b1mpu37/DKJz7OzTdeY3XrFsSePZl9Su3QKTiFcRqpKsRyMBeEAzDy83C9sIXpkGls/GYbfeY5g69snl7yQAtfnbacb844uXmHX/n8H7K4dp2w7gg+kufEgweX7PY7cknkueDEon+Wq4Gjo2t03UAIPbEDxFJIFMM0BBOvetdZl+4rotqA6gP2ooZpSGP7+kh2Dokr4vF11teuc+eTn+Ne2TI+GtnvL1kOLSyymvF/jA6tiVrzle2FiBJiK84tEKNQ8EvHSnqmeWaIES0zvlY8lm2fy45hUNZHHSI94+iJ/cDlZSJ2idUqkMeEl45cMm7RmMhFKVUZ+oH9OFOBcZqNuNeU7DkpBMh7Y0QH79ECZDOds40PLdHDKOHOORsFpTQHhoJrDGbzCwGwSCcfPHNKtokVcwyojYNTcyGGaPYi1fy1F1rZPrjP+Pg+UYBcqEEtLDLaz5/2M1mV7cUOXax5cPcMP2a6GyuT+2DpyyWPXO4uSfPIbme6v5TB+wnnHON+x7R3ts1bdWb1GzrcGPC0OCtaW68QnTHgqwPvrDAEceRih5POoLWwnbY85gLtPHQBiYH1csFJN7BYLsz2N4g5EGSlpMRHT57ywQ9+gFv0nN65w+033uDVN99kdXoCPqBdYFfUnpsi1vUeRufGRW8AyZVrx4t4vbiFyfwz7N5VuxHWwzEhmLxke1HZbi8gRm7cPuHu3XvUB3cZdSZnwUmk75ccn56wXq946eYNhn5FjPGKTQ00QqS0h8s4OCrNZkIKznXGnFU1nlSzmFDvsaW0GTV5bRIRqYgH0Uwql7z0yjX2D9fsHhcUGPeGMzkHwQteAikluq4zjMYLuc6EwWLCHWLr4lqQwRtHyAkudKhW8NleuyTmvMOHyNFRYL/NdF3P5rKyWvv2s6wI5jlBGx9EjMdDsdSX2DkKhulpbexnTCuXshqPytl7UYrp3mrV5hQZUDULkkoBJxQtNppKMHqDuNYRPQuIjMGTtVJy4z9VRSQQAhSjSDf9nz1WNWWcK3QoEVuxp3rAuQRKSz0uBdc7pnFDiMJvf/FzfPTOT9GyQ1VIs6DimDP88McfcrQITHvjS/dDh/cWxZ1KwnmHzso07unXHtcr5IrGSnbPDhpNhehMuFtVyOUKfmosb6GqgFemWtAhWujofobtzHQ+cY5HqXS9Z7nqGVY9i2VvLhY+ErqASzNPtxfcfefbfLPznN58iVuvvc6rn/w065de5nhY4+hIznysnEjziZNGXD54j7+Y14tbmDD+jIgxcjVXXA1EGZDmdDiOMx44e/qUb33n67z08m1u3LrBm2++xum1WwzLFf1i0YgHEWn/3ENqqmuG77biDs88bjBF/oEseMiIFxdRZ77WVQKJDi8Fz4QjQ5kRLezPL7j33tu8/c2v8OBH32apO5Zq2AVZ2W5NohC8MO0Ku22iC8Lx7QWd9xbtgxgLXCqL1cLG2SkRRekXPWUqqEZygcpEdJ7FqiNNSvA9+82Oy6fn3LjujWg3F0oHPgYLuMwma3HeXAe8WGBo80e0sW2u+GjujhQbw7xvvkBt7FQ1CY10jmmyDiV2HSXbWn9KJuQ1gzoh5dIIob691/VnPvWDKZoCwXuC84zjZFSBCjH2tiywP4xmZ6/JG0YmwdJoUKXrPWhl2QlTuiSyZX3UM+5nzs82aJkYFgPeR1aLjjTtKa0DXC5X5DIzz2ZdG6Mnlh5Nk6WvdJl0VOhfu4lIoWwLacLY2eqoUyVUT6BStTTQvIJXJMCOTHXCXCYUh71qse612ZnsdxP7/Z76yKLEu27BcnFE1zmW647FOiAukXaJe0+f8N7bb/PNv/kb1jde4hOf+hy//Pnf4M61mxwNA97o/xYu+nNwvbCFybg5JqkwLVFlv92x21ZqHxnWHbdfe5m3PvsxPvWZj/Pmx17nxo1TYoiI9xR1VGmiUTHbD1F3VZQOXkO2qfDUkq+sQrzzVLV+KHhb1bZMYMC2ak48viqdFFzasJCJJw9+yjt//23e/s43ubz3Hr7sWYXEohM6byd9wYDfNJm7ZJCMJ3CZMlVgMXWsTo7J+wy5cnq6IopnzBsWvRLcCApzTRZWMDhSESSAc0pQqCnjg8VPe/F451gtV3Q9lJqpzIh3OBxOxcYkrcxqCTCi7VEJQinGIsYLIXjmXbHtTz1EwcI0zdYxDT21FlKtiHNMKZGrEpxJKIL31DpTnQlQgUZGLVefiaqFU9pnYwWS5jmkarozSx+xvLQ554Npoy1IUsEDITiEYCy44vClsHm8Q6Tig3Lj1prHD/ZcnO8Yho5F6Fn6BSOZrJVFUMaU6KJDZOD8cWV64hh8z+lLHXNV5sWS3/5v/jUv3/lVPvjxR7z3vR/z/nff4endB6TthrCv5uOUxkaCBa9CzYVVEIoKUYWCCXxNpCuU5/RzZoAXyDlbavBuMkhBlOW6Jw6BxWpBvwys+p58OXOxucfXfvKQH3ztO3z+87/KxeMzPEJWrt73w3v/ol4vbGFyz52k0uxuVydLXv/YK3zyl3+JX/jMW7zy5iscX1+acn+I1JLaJqnifHN59L7pvZ55LYcQrh4GMCzH+2cd00EgabHK2mxzFR98kwnMOGCtiZB2fPdrX+bxB+/w4+9/m/PHDyFnfJ4Yek+HefpoK4hyoEFUKJO5Tzpx+BK4uFd5fHfHXC8IvSN2cOsm3Lx5TH+0pu+VIQ7gRnMK6BO5zpRZcAFUM75z1DJxdM2TZ8X3ivhKqhVNlmYiXihzNVuUEFBnsgX791VqNidGxW5k551p9jjohJQQI/NcrrpJMKth7yOKELrAvK8mo6mljcrVfMuLWQ07cVTMysMSeg3zAhuNUsoUx5XV7GEfok2qc3kxsRg8wxBJNRmg46BKIalDEwR6dOrZ7nacnz2hW82sjzti9Fy/ueLs4ZacE+dnMyfDwOA6ZtkhZcZJYUqZy03h8SPH47szn3rzJt4lxGWy67jxuV/l1c/8Lq9/seO3k7A/33Lvvfd4+5vf5N2/+xbv/+Bttg/v4ccdy1JhSniFOBcKShRHdR4VRwIyitNyFT9eMatuwTNrpqiB/7kI50+zeTz5gnQXDIvIMCyJ/UC3XHE5n/OVP/8KP3rnh/ZNms9Kw+j/sTD9w64MdvZRVPnML32a//5/+G957WM3WRyfmr5ssMRRrTNzTlAL1Xkbw5zgqp20YLiJiDQOSJODNKzpyr4UM2jTSovc4er3VY0l7B0EzUQy+uSnfP1v/yNf/4s/Yz5/iCuZ/mAL4j0u2aZLgbEUE7diJD4pHua2bQoRzZXB9ZAru8uZGmCvme2Hj/iJe4g4YX0UuHGz5/rtnmu3TxC/xYfZXivFXAd8tey1PuIlUlOixkJCjSMUPEGFXBKlVoIq82TWGM4L6iohBubJ5CY+OILvmecdzkeGhSPNc/Mlt/HAVPhtO6dqdIBkndU0zSYoPjC3tXUzKmgxo3/FLIa11mf+1q4JkZtjo4HqLU9PlVwqi8XQYtcdLgQS2UbVUuiiYy6w3WYLLJgTKRduBHM+yK6gLnDj9jElF3aXW+aWCqzeU13PXALbDI82M/d2cE7ks3deZl7cY18q3eoOL7/xGdQvqK6jBke3XPPmnVt8/Dd/nXSx5ezuh3z3b/6ad/7mq3zw999l9+ARebNnmA3Ydy1ggFrxYsIfJwc7k0NCTQGsY02iZpKIA4kU56hq1IwxFfZPzol+C/4pfjkQFj3zbm92NVX/0V3g//9VAIfalM6N2zf5hV98A4k7cDuqC3YSqyW2moVpG98OlqbuwNRt37GUq+9+MHnXZvymB1dGNQEsmBdzFfPVBnBekJoImrj/wY/5yz/9X3j0wY/w8yVdSW2b56mWzYurEEJHLRltftFoizIaFU2Cp0NLIDqPkHCS6dadqdNzwAlWcJOSHyr3Hu756AcbXv3UCa9/5hrEkRBGfDCQOOUJ3yneWSgBneKjp5TZRgYsJThE30SgBup68UY4FIt6EufMZzsVG3N9YJwSuVRjhiuAwzthHE3xXsbJxjtxhK6n5hlRRYt9z4OpmTtsWLXigkOCXKnfD8XHHSxCVCm5tsIEIhZAulgubAVfM3OZqK7iOkehUpyQnK3533rrTXabLQ/u3eVodUyoO7ROzNOM75WC0i07wmKgqqW5aF2wY8kmR86nyOP9yPHrn8TdXvAf3/ker18fWayO+OJv/Utk7Mj1Et+vkK6nuEqVYhq7k2NuH6+4+earfOH3f4ftw/u8//23+dbffo33v/8DHnzwEbrdM1SPH2d8sUPFudKEWCbWq2K0ClVHUEdJs1nOMKNeyFRyNV6aFHDZttfzPLG7tKw7I97os2Uo/9gx/YMuU3xVss4gheqrcXqCkGtqNg+2SVMiufFqTN0fUFfaQnRGpEMkNBzJ/czJCwf8orYxohWkWtGqVF8oEiD09iCVkbs//hb/15/+r6T77+Fzoewy5prWLFGlsq/GUSoCfbRUWEHIs1JnQVNGZ3BDREvG1ULJI651VdGb949TqIdYJBVKEZIK9969YN4L3VHk9hvHdEdC5oIuKrFL4CybDtWGJ/WMM0wYJpPVOjjVStUMetgYeWoyR0vnhBAdzhuYHCWi44wP5q5oLHbBd9E2TdVfGa2lKTGEAclio2x1+PZeC4KGbLhQExmXFuwg4s0RshY6dXhxFHIz6oeUlK4H1QnnbEsZQ2SsiTFVCA7pOwgdb37yFzhaHbHZXzKmHYtcWASlakGqUGaHxiWPNhkJPcOwoL9zjVED2p2i48DLH7vNW9ff4pe/8C+4ON/xP/9P/yNf/sHb/Nv/6k/4+K/8IXsRqCNMM5K96QyDM82fi0xaSXXP1Ff6N17mcx97lU//we+wefKY93/wQ97+u2/w3je+x9Mf/hTO97h5xjPixYrUwXSwKKh4qjgIShUTWzv19OrIohSqGeaJUEj4qlAiXVErmECopuHMSGPIv5jXC1uYDrsf2jpeGoBqu3gDxGtRwJsPtQ8U5YoMKa0NdmI+QE6beVezC7HT92CwlVtCaiW3Ue/QZYnuzVqkZEQLu4v7/OW//9/Znv0UP20gG5O6lLb+b1woPAx9YLmMBKfEEMjzbEUxqumzBFKdiFdExop4ZxhAcymUqzGtELuIrx6SbQqf3L0k3avc+8izXDtu3um48+opocumdSuFPFV2szIMa8QJqezoluDcbGNStjFK1Ez251zpYmzx3IbJuSh0Xcc0TiiRKVe6GKwApUy3iIxTausBY7OXbNbBsQHaWVPznsq2dQqCRG/6OCxHzYsjBk+tmVTS4XRqHClvJMZaUPWUkpoBP+a4gDkBiEItBXWV/X7PvXsPuffhPQKeXVJc7EglcLGvjHXgfHac7xyLozV/9Cf/ljc+9Sn80Sn4FapLqD2Zgeoi6xsd/+yf/xf88N1P8q/+6/+OGjzFWYduhlPZ8MlJydm2q1pNwlKc4htAL31geecGv/zyLX7li18gnW24970f8f2/+jrf/9rXePTBu0z7Pb4kdMoE8c2p1HybqmtdvDTqgQpeq4WTIu1zyMQYDfhvtBuVZwTLF51q+UIXJuVZ5Iw2G1Hfft85byGOahu3KeUr32Owk8tcCLNhRz62jZxtfA4BAM///8EsPpfWVeREmh+R25q6i4F3v/fX7C/uEWSmeZnS9UIURwielqhDLsU2ejUZO7yOKBYXXsaMIaCQNBF8BIIpw6uFA3h11FKuvHSkt07CiUOk0kWPR+nEkafIdj+TzndcfDjTd7ZpWwwrdheFR08uKfKUsBSObgRWp45uHQmx0nnB10wM4CQTXDHpjzcT/xCNc+OcI19m9upw0bhDLluisHfQqV7ZsNi6X0lzbWNapesxEqhArY6kghaHC+aF7Wo26gGCOtN+hc4OGecGphlymul6M3UTZ6t3BwzBBp/xoCHzwpx2/PiHb1M04PpAVeFsKpyHnrOtMMeX+MXf+H3evH6Hl15+mZu3brE8uU5yMDqHuoiTDs3N9L8aOP9P/+AP+O0vfQntOtuiQXuNhigfOvLDPSxwZSFTtV6FQ1SgOMEPkXDnOh+7cZ23fus3+NKjf8OH777N3/3FV3jn69/g0Xs/JcyVUIQwF/pScEVxElEnTMAk1glJqY0jBUkwLaQcwqV+vq4XtjBdXU0YqdoMr0ppJ5RcMZa12bUebohDwoVvefSqytxywA7FCLjaylkBa52Smkd31cJm94S8v0vNiRlhqoUff/er6HROHxTpI2FwUAqDD+bPLVBKJc1KyjPeQ8lmkoZT0myF0XUeyVBrJNGhORvGkCsxejv1m7GXrezBB4i+0nUBBVJKlKoE11Grx41QpsK+wmaeufSJebT0lFELyRUefrgnrqA/MjrA6UnPtdOOk1MHYcOwjKQgiK8UPyIRM9CPjuV1RZaOOkPZVhgFLUJKikiHqlEuSi4gAR894gNOilmvzCBxYHMx8+RyolsOdIOwWiSqK9YtasH5YoGhvTLnSp4nww59QYLY7/ee2Hm0GFYl4qgJfOdwS0+vMCalOE+VSOgX9HTsumt87j/7J3zuC3/M+tYnKM46P62JSUaKFgjtwKojIThy9njXm2PCamWUBhGT6SB2GKk2C1vjZqmao6SKNEfT0m5mwz9VoGhBoyM5JS88UhVZvcSrr1/njd/5Att7j/jJt7/LN7/8V7zzjW+zu/eIupvoUqUvSijG33bNMbSqNj96xXuY1DhUXhV3WJ7+7KP1wl4veGEyaqR9yILzzrY3og1LahKSBla7FiZpGFJBXHOTNCo3oJRy4MjI1Z+t1TyRDtslMHxmv99Q9k9xxTLV3vve99g8/ICT5YJtmlCxzmqIodEb8tVWKQS9Ync7ZxQBFVPUe9chnUOkY7W+yZO7T6gqdHmmc6as9y5cWfZ6cdbpFZOmiGsYM5WI4MreilipBDF+UgXksC0LihQ4Xq7JzoDo+UK4TMrjn+45PoHX3lxy+5UTisvEIyhuBjUCa4jG6+pjTxcCeRTmDPPo0BxJo5JTZcxj4yI5FAtZ6GKHAvO+cn4+8vQ8cblTzi6VWffceGng+rXM0WJJVSV6EFeIUZg00a0WJCaKFkLw+D6g4pmw4heHiNZsMhcf0NCRxCw9C/0AACAASURBVCxYak34rqIu0x05uhh49eOf4lf+ye/gl0v2PDDumixQjXg1FwRJzvSQmMGarexNs4mY40LV2hJR7Da1psqcIg5bX6iHiZxDWnMTNBDE4q9KzqhAFjMPxDmSD0jwxDde4Zdef41f/tLvcvbRXf7+r/+W7/7FV/ng22+Tzi5ZzIWQCy5NYItohuBRMhFTM0Vv1GK0LRd4Ts77/1GsXpTrBS9MdinGQfLOozobl0iNUO98y+2iMbWdg2oGarm0iBtbCT37fqoGyDbeUtVqS0AcOZvAVNVCC6eUGKSyefyQj370LjIl3LBiGgtd20o5Z4S3WqqB8k4IIogLzxnF0wzehFyKyThcZXmyJk2VJz99gFel5oaLiaWoeHGoE4LaehlvEUBga3dN1mV5Mb1WC1rBCeSa0Jb2u+wcoTOrk4rFOvXDgmlypGnkR9/f8OjuntPrAy99bMn6+op+Keh+JtSOlJWpOMZN5uz+ls2Twu5cmXcjl5dGPbh155jj0xMuL7c8fnzJdpdYr01EPc6Z7Taz3cNYPJN6tnPlJ/cvuP1S5MaJsuwDx+sO1cmcHU5OiUAYlFIv6YcF0HF5vmPOM94rXSf03YJhuaCoZ7ufrEvwRikIVcEVduMFR6uIl3N2j7/Nan6fcXzE5TxzcuOTDItPk8preOmRaj7nVnQqXgrOqbkdtM2WHHRmbbNr3XlC9VCUDpIqveLQHegqzl9VK0RNg0kuqAOKEEpsmKmycyA+sHzrFX7zY3/EF/7LP+T+D97nO//+K7z7l3/L0/fex42K5ISvQiiOoN5GuAqpGOnW8dzW8wUtRs9fPxeF6RkQfmCuGhfb1u80nodDWr9qZL02pnEY71qiRAO8D1/b/7fZHGk6Lk8pwtCvWVx/iXTxiO04snlyzrLrOX9ySU5Cf7AnKaP5AB3sWas5Hh5+ViqVUgStrmFGijoPZJ5uH+Oiw/cKk2nBzJLJSI4VGy2cO8SGY0BxMFDYeceCBTllajHfbHFC1RkQs1pxIJIpOeG80gWhxMRO94hWyA5Pz/mDysWjmbPHicVpZH0cWB117PdbLi9n8+8+Vx58tEFKRGugFqgSqRQuLy/AXRoBUh0pCU/ORnsPtJAyljdXPcVHfO9IU+GDD5SH9xQvO157bcVi1VN15v7DgneJ9YmBveN0QdctuX93Y4ENUfFBOTldMOdzLrYzOOhPAkfrjqELyDzhQiUuPHp+RpLvkp9+wOmxoOEMv17xZP8et1731MV1RHpEOygWrV4BXxVXKw4DnoWDfWFrOho8YFvFAwXFilerY8/u5RaPVQ7fQ+3ztk2hwZYxK9VXcvAkrcbqj4I6CCcrXvn1X+LNz3yK6V//MT/62t/xnS9/mY+++12msw1+Apcsuw7sYItiG1F3AL0O1wvMaXphC5O9h60FxeJ1qkoba1IDxV0rPELVbH7XzdS+Nt6Mw8ag4ENrncuVlaqNgsUCE0UBsy61GybiCIT4Oove8e77XyGEQvWZXd4zlUTXV4IToj+kUGgrSs3yohHacoZczAs7F0W8oKEgfk+SB/QxcnS7kp1j3sE4Qsdg+LhTyAkotiXzjT7QpA1eHNmZ85OIIuJbYbCT0XlF4rNYaMWRCdTZcDWdKjoZ7yZUT54Klx8qu3uVR25nD4gKSGRKEzIfoogqQjZRr9qnUav9jKw2ktQirXs1wFtLNUAW8/8uIgTfQXDsk7k/nk/3WK0DLhTLYqOwWkecqzgfCB2cj57NPiEhETul31ROVsfM+5nlolJSYRyNUBn7SpBCTZW+7snbiW09w6EsrwXyPBBiT5o7Qj/QyF0g5rftsW3WwdJEtM1EbTSnjW3uqjOuQMMHnR2M0u5obWxu70wGZW+MNpWBv/Idm4MtD6gV18IKrAv3dsA40DX45U0+/cYf8sn//J9y790f8p0vf5Xv/flX2f/kQ4axEhOEXFmJ4xhhrxYd7it0dh7xTM7+Yl0vbGGC51aaB7rqYZ3vbJ1fGgVA5ZlO+kCchJYeS1Onq7a4IyWl0giV9lNqw6QcjToALYNLqSVzdvc+Dx48wfmBy93MfsbsUCIQBCjUWnBan1vh0qKroVRHyRXBuqTYOUIvOA998MQiuEVE12v255VH758Za30CJku7qM2nqUq1u0nBq4VhppxNpKmGsU7JVvI+Gq/LYeZzhUrNFS320ImCU9OypeY0oLlYMU8FvJqPOhX1hVoPBcjinbSdytDsTpxnLgUJkTRnE7OmQm3HRy7V8s5E0ZoQ75nTnpTb2B2V2AsSjdukDsQ7xjQTgrDoeo6vXWfMT9hOG5Qm9dEFm9HTdwMa9yZJqcbFqloJwRGDbQVLnUAim30mdSuun7zCmx//PVz3KWYx4fhB4A1c4ZCH++rQBR8wSrnCLo0kar8tDSIoFrP0XHcOXOkwtdiW2Tl/tXm2/AqPVrtHUePTKY7S3Ccq2iLDWw7eyZJXfu1zvP7pT/F7f/InvPuVv+Ibf/5/8+Hb76GTMD64sFw/FJwcVEUv9PXCFiZ9/r8GFF1hRaVqC/RrAKJWSzkt5iOdmnKexmsScTbmYDeL3Uy2Qcs5kcvcwgIPGHm12V8gpy13793DxyMuLhMXG6XrTyh5ZnG8xJXM0eDZnZ9R6tSKk/1cHzxcdRR2Yw6LSIiFOID3ShSL7haBOXhyLrzymZe5//37pikrHZI94i2mSaW2qtkM+lMF8Thv0pBa1Xyxg6dKsJigYnhYrWpOAAg51ee8vu29KanZ4M6VWiFTKd62TyL22oqpJywUoZrOzyKCtBUn17SN7YFUgQqpNHsWZ5yefOh1BUIH3cI2q74TYm9bJufBi3Wh4oQ5J7a7Hf1iwWJO5OIYx8KuOGIUrt8c6LoEfcH1Ht8ZBtVFoQ8CueI6YZqMHb73ntdOPoPvPsWst2k7FA4jPfqzxiDPUwEOxcY5udL3HSRM9msm8Tl0Us9DCHLwxRV7//W5n/N8jLe09914ey1Xz7X7q3lASbD3vwaHuoAbbvFLb/wxn/1Xv8cPv/Fd/vLP/opv/7s/ZzvtTWqjahl3As/20y/e9QIXJrm6cQ+/AgUnUKQxM0Qa05irwlLmGSpoE+MaSF4Nj6qgKm0zZwB4rYVcJ7yXqzFMxIS6tSbOz+/y8ssvwWbL/Y/2bDae/NQIho9XiTs3r/HK6y9z1n/Ag7vvUaUakU4L3hknqii4zhi5MTq6weNDJXg1Ylw1Qa2WQLdwBM2s7/Ts7o5st5mgAZ8qLgq1GCa2GCxUU7OBvInZULcYG+2wRUkJIM7SYb11SDUXcjIjfq2mUctzpRZH1oqvtsk0vZb5hJdSyCilRhBHLhZzbjQOQby/8vGuqV6RJmtuLPsc2+aobU4pOAksFwGiEUidg0oiBGcAMRXXUkAOVI7tbkOIA10AVcd2M/H4LCNuYjPBW59ZEtyOKuYVPywWV92SVKHsA1o8tav0/hZzvcZYjyk6IDK2rudQbBqQLYfk2+c7Kf0ZLeXhKL0i7WpGi0VBHbZ0zyLh2ywgxlVzh6LVvpfwnxayZ183eU/DsqpYEIR3diik6JipdDeP+eTv/jYf/9Uv8u7Dc77/7/4PcrFxtDX0/1iY/iGXPjed22duYxW1gDOCnyV5yNXGrQvRTvViN4U0R0ga4/sQHnkY6cxr2zY4zgMOnK/UMjPN50aKLGfc/fADzp/uuDgX7n4003VLulBJ9RIpCzo5x+MpzpJ44yIwjlsyhRpAvP1LglNc0KZeMY8elcbDCgJpxjvPlC/or0cqA/uHicunE8vkcA1Do1ZCAV+Nl1VSag+QI+Vqbb5vo3AVXMQCMtUsRaSa7W2tjnmu1HoIveyo2TpQrZCqR4snqVCdkrUwThUXPKqeXMyMTZzHqdmcqBhBMquB8YeuqqqDqkjN+Og4Wq/RAOoLoTf3BqS0120Pqw8O02NbN1WKMk87dvsNi+XAuu+ZU8eTp3tSijx+tOfW00Ac1owlkcZEmTIv37oJqjw93+LdKRf7HR88OeM3fv8tcj1mPydwE10oh2XZc0WkXomLn79EnkmXDkv4QxErtVx1VzkbA/ugyzx8H9+8wcU3D0l5tuFDf7bYNZK3jeRqo1w9vE4M6HaqlIZtVVEzrysVFzv0eMnsBF+EgIVCVHmBH35e4NemtpN97quD8voZ6K1NgZ1aEm3NyQBDKa1rsIw0u3cOmIC/oggcSGlzmSlZyXNinidqHallCzIz7e4z7s64++ETfvLeYx6feVYrIU07lucz82bDT959zJ2XlqyPB7pFJIkxevsuMO62lDTR94EuGuiu1Qz9VQsqnlydPdDR3CmymLwjngRETQ4zPslEL1Sx1fRcFV9AChTX2MmY17UEf7U4mKeEq4JEh4TQ3g+zyq1Z27bQk0o14p865sZozlWQDKlWE8dqJbf4JAXwQlH7e16cdUnVRo1cLD7q4BTgfCUGM+YLUXBxQmLL2KtGHHXeGPMcilH7XlUV0WyPf2ldc9mzWPWsjjrWS8fZmSI58JMfbNmcO1arYHKYMjFeVHYXOx49PEf0Eg0Dj/eZlz4On/2NgcuLHeKfEnzGe4f3Ae8tlt2y5HxbKjwrLM/jS4eidEjOPeBPYOPf8yqDw9/PNWOOUfbrBqC7K95dbRiTYVwHft0zmoweeE/qEK+GA6r5hAnFhN9eKKEyeROjxypXGMmha3pRmQMvbmE6gERyiHGy9IxaD0XnQJAszRbbWUdlqLVxWYJJULSRL5F61S6rVvbjjpwTBVtpT/vJEnTrDtUNpVxy/8N3uHwy8uRsS5qFNHuejiNH6zVp7/jg/S3UPffvXXD9ZuD0ZuDWrTXH6xWx6yzlYzFw4+Ypy6EDKVycP2Xab6Dlq5ViD2KOwpgm8BVNBe8K/dDR3xjYRwsoULFCMNWKV6ySeTVJIQJeDItICZwY6J8MAK/FAjjBk1Mxs7raKAhGoKaWQnJCdgLeOqmKoGLulct1z7Do8NFO3t04MuXCcr3i4rJwfrkh5YwLjj50Ror0Dh8LfR/wzgBuJVsWmgRUjRSY63Ql6bEu2KgXwT/DE0NnoZQEGBaO88uZOc0Ed4QmuHwyMU+FxaCEUHAo5w/PiGHAc4oyMQzXEAm89Ynf5Gj1KmmGKjsmmfDeE0Ow7D5vpoDeBUKIrUA1n6+rkbQ+G71EG7BtrG/whjWJaz7n2sjCzeKlWQwfOiZphcm7Z4/lAY+6+lktKsqcVLXZ9goZB2JupQZnKuqNZlFb0TT6ws+gty/s9cIWpiIBR0I0NONRwxFyVZSprfz1WUtdDozvdvrgGobUQNZQqaQWhlioORuOpJmdWkSzS4pMI6THTNv3uHjyHvHiIWFecHG+J5U1Ryc30GqpKqnAk4s9XVhysZ243Dh++n7i6HjP8XFifbTn+o3AjesD8TxwuVFu3roFvkNDh/rEPG2pmm2D1DAwiqMW85suYcafOBZHHUd+aGBzYZ6VLnYWMxQ805zwLkKtjLuZMhemuZBKoypYM2nYUqm2xq/2b8i5BT+0P5RqpDjB+UTsldPTgePTgcXSs1gFQnCUYrHtuymwn4P5XcVD/p/1t+IFXCVGhw+RfnCIzBbm2TLobNo0nrWv3sB1PVifWIccolkCxy4gvoBkqhRwyvnlnn3G7uTqcGVNnRW6nnncE3xhlkAIC0vtXVW0D3zsY5/lrc9/kbmuKZqo7I3RXR3juLPRyVmX1HUd3gf6vr8qUFUF54PZxRS1v4sal04yWjNeAyKRRGnazYLH4bQQtGFULfDh0MKoWjd1EJt7b4sN66AaxFEqWg5AUeNQtWWOUqhOEY04NR1joJhfvbPFiWDxYyIvbs/0whamw3UACQ+CXIMqsnFkzPjBANeqrVg14Wezzq1iCa/TODPP6crcvuRk0UkoZCWnCeYJly7ZX5xx/vgRJRUqS7b0zC7TrdeMO0g5QfXsRkVlML8ccewn86Xe7ic2m0rVPUdHwku31vRxh4hwuXmPxcpz69YR104XCNE8ixRCUMo8kacZphlvNni2mepmasis1kuOlkd430zEsnUb69LZhqc6TkowoL/AOFXm2XzF9/vMblsYRyXNQs5iYZ7BfL2lOoILdEOlWxZOr61YH3Uslg5xCecLKvt2aic6DzWCZGG7H4mD59qNwDgVUqnMOSPeGdAfQ3OHaGZ7LcpcvFJ9bhqv5mbQHhrvhb73dEMgdI6qhk+V4thNgYcPtpw9HqllQEiIc/TRsVwtcV5BBmpNZIExw+rkmM14gTjHP/+9P6RfHZG9o2riwNCuLYDBtm4JUOY5NZlRIITQitSADwHvl9bhKATn0Fpw1TbE3o3NE8xkQkpsmKegzM+2fFckTHcFjh+851UrOdeWOGMWvIcQB3PEKK0r861rs/FXnKM2F4or4Ow/uV7crukFLkwH3sfBgsS2cKXM4MxdEW2gKp4QAqrtpMawijkVcq2kXJjTROw68MI0Toz7vXFEMCyklhnKDskbNpsLnF+yWh5TFwuuvXmDrT7mP/zZ9xinzBAG5pRxvsfHypwnvERSdYhEBNjuLCNtGoXtJRwfrdhe7tiOggszoXvEyXGk6xyrxZLVas3TJ2f0sXCy8hyvBqLb0QcDh6VTfOdhMTP5GciIU/xCjPlezRZDGy1CqzHgw0KJEhgy1BxR7UG9JZ5URy6Omj0lgVNvALvb4UImxITz5oflnK3HK5UYhNAJoSqalbRXYm8PVzc4+hyY5ko+dKfNNLYUiNG1bdahEFWouYG6nthcNH0wsXIMEPuMiwJiibkpB/bbjvt3z5lHR/AGLseAWd7WPX3XE2KPuJ6Ly0syletxQS2eV994jV/8td9Cu4FSMtXFtvQ1PpU20mPOTdjdCLQ5Z2pVQvDsthu6LhKHC7pe6OOMY0LznjQnFsMSqqdzS1R7KgO5RopEcHavFm35hDy/dftZ80IaF8tkL1xREwz7cldFzA5oww9rLThRXHVXf/7n7XphC5PSRLrVtmulmCOfG6xjCq7RCRA7SUSMwCdCzpkp2SncLxYcH61BhHHcc/b4jM3FOblZUXhxeDVWsgN86Lhx62Wic8TQsVv0dOEaHxtv8+X/8H0IldMbx5yfnTOP5jlUNVG0slqtGcexxXKD6sBurPT9dfbjgqeXBecj07RHp4lxMv+mO3dOuPtgy/27e6LPvHy745NvrTk+CrhlJSwrQwj0fTSRqKs4X9HDFmtqZMICXWybJDGTOQVqnZHqoDhEgrX9ruJ9oVTHPM2UGWqqkNV0et7Y8JbvpkbIrIpTaUnBRqr04vClEp0RDEtpViehBVgWwwKLVqNkoM2ET3C+cmA/h2DYkQsYZuaULpi7kOH4kZwCm8vAdiN88OGOs8eF6I9wfsBecLYorpqYM4Qu4GPHsFpaVl5OFDfwO1/6F/TLE5L3eFeMPlB31JIa1622DjYaN6w0C502+dRULJ5rrMi+0vd7vNxD80+YtnfZ7zYM3ZIh3iHPwmpxixu3PkG3fAWJ1ymyMKa3GvLjoZGGD4k8XNmoqB50dgBGdbHfOwDj2hj/tgBRpHXOz7hRehgbrx6utu3jRR3kXuDCRGPBHj4c4IpZ68S1pNVoH5Jabv3hlIsxcnx6He8jw2LJbr/nwb37PH70iM3mktzW6yULvusJrsO5aMB4FcR14ANZDFCs1TyQxBWOjpeM897W5NXTd117fYX1qiMGx3472RjYNBqpKE8fPsb7aKf7nKgUptlO4oePtlxebJnSgjklHp1XXpt7TrvI6c3I9WueuitcOzllnhObzVOoCWRizrPFK6VKbJ28ajH+T3SW66ZCSYAoTiveNaZT4yNZPp0SgyN6B0GIg6dqZp4LliJjAmfU7Gydt42SqCf7Qp2VqVjSiwRHP5ijp9oWHFW5IoEWdYTOIV7wYgm7PhhGJUHI1bZg4zYz7g0nS1m4uMxcXCiPH01cXmZgIITBeFQB+sWSUmdKrniEcZ7oRFFNLFcDKV+g3ZJX37xO6DNVhK4EvAyN8CmklK4IkrUFc7aagIgwp4Sn4KWgecLPjjRekvQnLLr3eOl4izsuzPMOrZekkjh/OHL24JTsXueTv/pHxOXrqBxxiJ6s2Ph6tYnWgw30AYdrnvPFDh/f+TYdHDaECgfnAxGGfrCIrGJe6c9fhwL1ohakw/XiFqYrCpO0L54ltPoQEW8nWpoLOWeWyyV9CMQumvdPKjx5+pTNBzu22x1lyuRpxmsGMX/uWu1hWSyPmGdPzpGsnZH7vFCl0rsCJXP55CHBJTTPhLik75dEXaJMFEZiJwx9x7Jz5PExXYiM44Q75NV5WCw7wPLGUhb62Ft0drY4otBZ13dxOfLDHz+h726x7NeU2bHqBj66Bw8eXJKLx7nKjZs36RdCrU9xmHuiBU1itrWihBY1pQ5LRknFCplAQshScAsrDDGKjU6D4GOi1sSyfRjzpAgdeTKKgfNiozJq9q5Z6JbC5nKyUcIZPSB27opLBqZ2Fy+2eRRlOXQMPuCi8aTm5EhpzZOzwuV5ZL8pXFxM7Gfl6cVMpWc3QSGCenJuyw018mlWh4on5cqwXFLyhPeZzs/kcskrr68IwyPzZfenwIIgvvHQOvOTagk6V0k6bbwq1QIStDa3yjqT5x3iHkD9CdHfJdY9q8Wa5AKlPsENM0f9xG4c2ZYFXs4Redm+rxyMDQ24OGz3nseYaOCEdVRGnM2JZndsWzpBGqWiUqUyN5JxkNC2ogd7xecerBf8emELk8j/+2vz9FbDt6mlkuZK1/WsVh3Oe6Zx5OzJIzabreVw5WyAd65EF+zBzal1WAknnpT2TGNE6dDqgWh5bX1k6B1d3LJ98pTp/B7HQ+Z82uHCMbEP9KUnxp5+OObpxUOCBHb7meg7uqFnP07mt+2VfvCM0wXOeUIQYre8wi0UmOdCFTBbWuHB/RGdHvKTd8+IAa5dW7Bar3j//bsshp5SJ+7cUd566zX6LqFpSw0F0RnfQR+j4UEuXIVbirelu0NRr1Rno09cihEdO9uAOcy9oA/2sMfgmWc7IuZ9IWezXvG1ktWx7jxTUtbXVhxtfePtPGPji3gjsoaI+B68oq7gvWXNRWx7t91X0iby+H7ixz8auTxz7DbKnBxzAh9X9Iue1bqy3ybmuZLyzGJ1RG0Jt7UY+3xYDPSxZ6rG/3IObpyueeW1gb/48p9y69U3efm1z/LxT/waKYGLgayCy/mKh+S9J+eME0fOmUKh5EYcLYrOI748YBrfZjl8gKZ71DlBP9OFY5DEvH+Kqzt8XbDsXiG4TMkjyhLjS1hnWaW03EMjlTa+pV2Nz4QGi8fSYhY7KAebHZFn3KpaS8OctBnUPaMdPPumL/b14hYmALW92+E0MaV+IQ6Oo5MVWj273ciT88dsLi+Z5qmR0pr3MoUYPNdO1vR9x7hLnD9WcppBE0X3lLpnEM/x0YLtDmoNLIYlfRdYDB5JG2o64+XlGb/+2Z5vvXPOZvcYrUfMeUKkp9OOabpgfdKzm3Ysjk7Zbi/oho55nogCuewRqayPjvh/2HuTX03S68zv904xftO9N6fKyqriIJISKVMC5dZgGbbRm94YXnptGP7HvPfKaMOA0LbaUne7qe7WQFoTq8gac7jzN0bEO3vxxs0i2/DKmyTgKKCyCsj8kPllxIn3nPM8v6euW5SscC5y2A/kXGiJUukCY2tblp2mqgQ+lpTg19cDensEteR+P6FkzZdfZj7/7BcsqiNPnmk+/OaautUsuxYnJMPJ8+bVAVMp1puaqoqYOtC2upxayGThCVVAVRnqhKgL3M4oQ84SI3V5KUSHlA65kkjVEGNitK6svZWnArI/sYygtMTUiuA1N1cTIQjqGrJIJVRBhzISwiASqKSYTprp1PLzn418/LHl8gqQHUrV9Ksldncs4lNLKaZG0i9a2m7Jbn/CKENVVYRYAH+10UgEORbpQ1ISR+TV5y+p39RcfX7L/YdHvvX8+5jqnCANQjiUd4VEkRLeu9LWxeIhFPGE1p44BnKo8PZIDq8x8iVdvaevNJWKpHggxnJKjn5CuICKEYRjOlyTmw1Cg5YrpGxnYzPzLK1QKYG54JQEoBIVnykuw3nt/0sv7zwXqZxnWuWDwBhmqcGM2H2ww/Au7+Te4cL0ttkGyEXlWteGRV/jomW/37PbHrA2EGMgxICSZYUuyNSm4uLigq5fMA4j2/trptOJnBxa5WIPMVUJf4y3vHr9GimWPLr4kIvzNavFghgmXn36Gdu7n+Gnr7g4U/zw++/x1z85MYyxMJfdgAsnur4FEk1b0TSG4ylS1zUpeIJzQNHFVKYpmNYkShAnDlLEqIqQAjEkshKM40hT9ZjKEFwo1MYplFW1adlsNsQYub2+5pAE3AlSnVhsGh6bx9zc3rHfCf7up5YUI+uN54MPV1w87ticlaLXLxqUttRtxNQDQp1AOYIolEipM1k6pHrIRijJsUolUgbdG0J8CBWHnG0RsmZTIp6SKFuxrUWpqqCHdUDVhVzgfYBc4U6Gu/uG//DjA6++rBmmHkoSHsZ0xFDe+N5ZUsz4kOkX52jdMJw8WldkIi4ci6O/ckijidnjo0U2EhdGGgnj5NkeRnzwbMc7/uDmv+L5N54gRYXPHqIssegPgsaUQGZs9qgEMQYinuAnXDjQpFu65ohUpxI6ii9ztKCIYaYrpDIuHdyRsHtNmhpEFagrR2WWaF0hpQYMQtSF+SSAlOfMQDn3eRnBQ/AoD986ZQ6bvrbQzE9PjLNH9Nekffvl6x0uTA9XEVFWlUJpuL65ZLQT4zgRQtFolClDeViapubRoyes1+dY57m7u2d/2DEdd0ynOwgBkWG1XDK6kbvrK/anV0yTZbF4iuDI2QZur15xc/UGYV8zHV7S6ANdW7M+f8ztteeT3ZYpdTxEVbdtj9KS5WrB4TgWK4KzLPueEApSpG0bmmaB85njccSODucCq9WanATDtEermO++KQAAIABJREFUMsw3xuC8o2+XNHXDOA7UWgOC4Bzelcy65eKc07Bjt0+MMdEfDfcnwZurzO2tY7/dIKTi5t7y6ZcDj54Yzi4UYDGV5MnThqfPNI+fNqzPDFINmDqQpUeLWPAjCLTskNQIeSpyDXLRgaVihUi5bNIEieg8IXt0pVg/alB1xXQ0SFWYwGXVXRExDKPm6lrw7//9jqs3G5w7g5yI8Ya+WaJFTUwRJQ0uT2VYbgybs4794UDdlmhtZMb6CSUTORmsL+wsZTLDdMV3fmPFH/2X36PpEpPbMtkjN/eJ//mf/4/89//D9zFtT1ISoVVZqcqyBAkxzNSF2VISyz3pwomUR6S4R6kdQkwoVQNlSD0OE01Vo8QsjkwgZcLZO3xqiTYzqhN1vaHrVjT1AqV6UjFXzk6FPKehyNkrFxGz+rt4+b5+Rh5EslLKeStXRh8xvNsno/+3650tTDmLt28JROZ4OvDy1RcEYfFz5DTMR9+c0FpxdnbGZrMhZ8nrN1c45xntVKQGfiD4LU1VlLMhJKKfMJVlrRNVnTjbwOn0BZ9/OpYMtZAQ04gInlpn6i5xcDc8f6/hzcuJu2tL36xpmo4YHckn9ocDPiqCn+ibFkKilpqqrsmiDIylqmZPmWS9XheAXPBURiNEYLHo6JryGc47zs6fcjycmMaBru8RKKaphB6mCMkrZK7ZXjvu7gfCpxYXM5OF6CUiG3KqCL7j6qXn/jpgmo6QAp/9/ES/9Lz3nuH8QnF2pnjxouPsHEQ/0S0dUnrIrjTVIpBlnuUEYo4GEogo8F6gZI2uDDmHAqmTgqrX2OiJPkOWZL/guO9481rxi08Hfvbxlus3UAnFatNzOpxgKBl8WZWAzUW/QIiEcxMpS6ybitcxCZwFITUptWVojcK5EaMmmjbw6Kngo28ZZPUV7Spy1kQSgW/IZ7x61ZH8DrN4TBCUUNOHviomEmWjqWSNEokURfH2yUxME0YOGGWRIhNDRiuN1pHaGPS8vs8IYpoRPXEiyiPB1wQC0Tn8eMD1Zyzax5hKUSxt6e0mUMxWl6Kpj/Pz8UA8yPOAu+CkvfdoqYuWLZfT1gNKJcNbksG7fr2zhelXRRaZlALOjUTlCv0wZcgKrRtWyxWbsw1SwH535HSyOJ9mv50kkQucTQiMtkQXMFWLT55aBXSSICT329eIXDOOJZe+0RVSlqQTmYpnKgRPXTd88OGay1cHYrJkqmKlcB7rHXG2rBgt6dq+BGci2Z+OmFbSLmrarmFIA1JmrJ3IydF2FXXTsF51jOOeqq5QRnN3d4e1nsViQdd2RO/fiuvGcYCkydEgksbQErwneIdIGnIgxRGBLh4wKmq1IroIBJRZMe5OfHaY+Hk4UtWCp2cd3/veUz76jQ3PP/K063tUtUfqkZTLTU9OJbdPGZSqCYAWPWRNypmuL8XXWcnoHCGPDOOAQnE8OF69jPzFj49stx0hPOWjb7zAW8t4GJnsvgQv+EwKjmwkWhc2kbVF9T8NGqjxoeipiobNEKInsCPlE6uV5MWHPR99e8NydSL4e2IoyGFJJPjI82ePaastpDuErsrJRM3hD6IEShRUMeQ5KVeEEZ0OiHxPrUe6BqqqOBGkLMPras7HS6loxWI2xFyB0CVFJllyElgbcRxx40RoE3XjkHVD0zZorZF6prSm9Bbh+6BLergHHjZ8DyftHDNFRa7ebkO/nnn//4Xp/+P1q19gTKHMEdyEmJM4+m7J40fPqesWO3nudlu8d8Uxn/Xc4DmkkshakURCG1fYNdmRpcOYSLTFmpFyIIXAdjeikNRaszARlIUg2G4d+1PFMGYWizP63jIcAsfjjqZtEOrrXLmSSw/BWlIspsucyk21299jjGGxaHCTJUVLVQm0gvVyUcSLIRCEpK4qJmtpu5aqqjidTnPGmcAYQ0wJnzQpSRIaomayDmTxtGUSuslonYpcIWWcdYjkWbaK9WrBNGoEG+6Ot2SveXOC3d0dP/vY8YPfXfDRd1sunhraPqDMVCifD8GcUbA629BdrHFjxtqA0JLJe7b7zGFXc3NV8w8/HUlesF6UOKmra4UQz6jrDSFIrq8npnFAZ+iaDUwn2rYny4wPEzlnptGSsyJhOByO5aRgNCENoCJNp1AG2kWxjHz0jRUpbtlsNLUp0V9hyiSTqRuJc0dG/zmH/d9z1i6B52/X74WllmdhpSdEjwsHpJgQaY9IN1T6nq62KGlRqpBApSzIYUQq8UxIhKwJyYBugRqBQVOMvcX2kkjTwBi2hCljuhYRW3JTkbRGm5I4LGVFzvI/KjJ5FsCmr0kGFILGHAj0a3m9u4VJiF85NAkEMSS0MSQBjzaPODt7RvSS+7sj02jLZjSXYEujNULK2TYSOe2vSDoTk8UYCdGW1sBnUig/P6UCJssxEZxjshPTccD4hLYGN1V8+enE69tbpE4suicsOg9YTuORyQakrEhCIWSF94nKtJADY7CYusxZurZmOA04b3He0i2aEmApimP/dDqWwatQMxpW05kG7zwpprKqTkVWEILAJ0HM4EPC1AJVNUVHVBnWVUPOlqqBYdgTogUNXaVY9pLFEqpqFlDGNQKDtwNIw91N4i//YuSTTyzvf0PzG7/5hOcftJjmiK7Ht2kb99sj1hZI3+gCMRu2+8T2fsFP/uqOn/zlwP7KoFXFcgmPnqxZrR8j1cj+cMRZw+mUWfUrFnXHsNsTspi5RoV4AIrKKIQs+OMoRpQOoBOrpWa16enWkvWmpu9apHRAKYZhipwOI4uVptWmJCGLSCMFo7vjzdVfs3n2EYLHZGZefJ7jm1KAFEhiJMkTKXsqJYATjTnQN57aSJSWUJqzkoYzUwWmmAhSIKqOpj9nSB0iqvKZwZOzR1CVF0scCUHi3UAYa0xbTk5t3xXdklRFZ/UgxqSc/ZTIM6gvzUbiB/HlPDz/ZRDd/Kve9ZPTO1uYxEMA14M6NonSjpiKiyePaJsVh73leHCFwpgVShmUEGSZQWqMaTE6M47XROtnvK5gDEfIlhT0bFw1SFOyvryzTLYMpXPMJC/Z7yN60vhTw/42sb8TCNNSpxUhXvHsueYHL55zc3/P55/vOZwgYXBjoFUKXWnE6LDRkgaLIJKiI/gJbQpPJ4lEDAHnJ7Quf24hixalqgzOhUKklJq6qoihmEuzNGXA6gs2V81vzZQSWgtOx5vijpeGHAUSBbLIKWKA3W5L0xpizrRdmUXlGPFTROQNbid5fX/PV58d+dnfvuK3/9MV3/vBhtU5KDMgZEYayzAe0UJjp5ph6Lm+rvmXf/qKz36RGQ49wrbUq47buyORmpu7HU+fP+VwCrz+ak9lWrwb2U4WO1n69Yq20tjhgFIg8CAnjPSYStIsNd0is9qU01FVr8lSInXGWsduuyNYR60lwy5zc2W5eNQgsqGVFVUesWFA5oC3n3J5+WPO3tug5SOC72Z4nn+rWYrJ4TgRvMdMlpYjXTNg9ERft0STcMlSenqJniGFQgtilqhuhTRrlO8wuUJhyMKXmZVQ5TSdJkIMoBQ2Vzhf4V2Lc4626wk+UjWAlGTKi6xgoItGKeVcTrIUWJ1WkhzeSjfnDoL5pT9v+97R650tTL98ZcBow8WjR2wer8hCc3+3JwbB48ePGUeHdwFnJ0xlkNqQksG7zHF/z2m4pBICnVusu0cI8O4EwhCjJweNEBKjDLKSBA9eeUJKJAVjzOyvBvJgGIcOw4q2eUaYJqJVnE4nLh4v6c/h2ftPuL3RfPKLI1fjkb2v2SzOkbQwjLS1wvvCf9K6xIprrajrmuA9Oce5nVAch7FwgYSikoYYUtHqOD9LJCIhldQSU9WkrHDBkXLC1Bo3WU6nE0opTqeI9x5EJOeAaAxOOhbLYl8IsWyOnC3n/6puSE4wHAMhLtBNy82riT/9Xwf+r78+8Ad/vOHDb/bU/URWA0kEoteMhyWf/6Lm3/yrr7i9UwyjQWSNrirablXEkCkh0Xz28S132y1SZVKyWJuQs4fMh4kJSGJEmokoJ5bnkaaVNK2k7hRdr1DKMY0jISz54os77rYDhDLbySHRd4ZKF7rpcPRc3g783u9d8OyppOkl5xcGyRV3l39GyAPvPfkjKvFtotwwBI3PiYghRUkKApEdcIOUl1RmR9OUF2FJ4inpOHlGnMRcAhqyqqjbM0xzQbIrsgPtFKhUoHxZkvAkpoLXkj0yW7JXWD/g3IS3FtM0NCnStItycgSULMwo5tZTziODB4bdg6bv12Tm/fZ6dwvTL+04BbBcLnjy5An7cctx8MQAL158A4FmHG9IKVC3GnIsgPpxIoVCOox+wnNEcYLsUDKiKkmIAc9EDDUpFntKFoKcS4+uDFgPpu2x+Y43b3Yc7msWi3OGw7GYYKNivx3J+cDmbMBbeO/pN1G6wvprru/26NAhpUJKA7NLXBsDJJx3OGcLLdFIrLXURqOpULqcjKx1RAKLfoHWBW/hhgB6TuNw5UZUUiFjxnqPkAU9rFVbNDUxkpIub0oh8N5hhaOOkRAmnHU0VU9MsczEYkUKEqkNtZHUfY3zcLJLvvz0wPXlNd/5Xstv/85jVhcbfBy4vYa//9s9P//HW07HHh80iTKAzcIxDhNa10hZE7PCnhQydUDAaINUAjudWC5rYhxI8sTjpw2bC4M0Dm0CSnsyFlO1iFxxc+m4fhPpusirL09Yn1FZo1VN21RIDF3bEPxADIHd3cA//HSi1Uu6RqGTpWJPinvS/cjl/iWn4T02j3+H5flv4mJP9iXOy9CA2CLyLW17QIoD1nsSCmGKEVxQ5s1y1hSllDBtT91vyLLHiJ4gIikOZCOLRzOLwgcjkDJEH5Eizds0j00e7wbk1FJ7SzNO1E2PMTXGlKVGpqCi00wNzfP8qgAK5gZO/PpIB97ZwvQwX3pIDo0pcnd7yxgHQhIoVXN5+WbmMgmEzHjnsdZiXQJqcpSkEApmzh+R1REYiTEgUASfEFlDLnlz1lqQGm8j5+dPiVHw0Ycf8tknX9GvgeoSjyNSdCuJGmMqumbN/n5ktRGY1vPoXHBx8S3WmzX/x7/6mMyBnCq0MgQ30Paq5ISZgrq1k+VwOFC3FcXtlsssLZW0ELLCKIWpK7z3OOcK85myIhaqECYfPF21MVg/gRA0ZgFJEGwxOT9IMHIETc1pD6Bp6o7oEzIVdb2uincu+JGqNiAiykQ0EkOPPSX+9q8mPv34NU+enTFME5dXjtNRIuUSIQzkWMy7aKRO5BAJ0ePdWHRdjcGNmZSK/0yYQNuOxHzD5pHkg290nF8kTH1CmYJ6AcU4FJLp9sbz5ivJabviJjgqc0a3gEWzRCuFtSPrVU9dCw4HV0B5U89XX+xY94az9Zqm2SLkSN9D6+5AjtT1K8LhK6jv6fgtXKgYvSNZj3AntNjRtROVAaEyIQVU1MWbqARKa6AQKlFQtT11tybEjiRqqhwJGLIX4BQ5SCQdwY2FJZZEiUonkrMjugmPJNka5x2jPtL3a5qmp10sSuERM+NKPvCdCttJPnhOf0kNnud/vcunqHe2MP3yJRBM48QwDCWpT8zY1SyJSeDn8Mbgy4NQNB2BQCKngEAjRUOKGil7lIgINFMcCNGVZNQc8cEhJXgbqFWNMBWEiFGaumvp1x23NzsSA5WR7A4Dj9fPWC8fc7i5Ij3r0fWJ/e5TJvcpF+sFf/SjNX/zN6+Z6Ehs8MkToyal4iAvkUvlv5tWEZxHIajaFm0Uznqq2hTYfAxvgWVCCJwtbVua/V0hRGIoa/wQPEpLbBzRRpGym6UNjpgmlDDkUGOnQNca/ByrroQp8giZUKoQFRARpSu0kkS1xTkNWRP8gv1WcHvnSXQgFuQkSAiEDAgZaU3NODhW/YphGNBSkaJD64wfHTJNVKomC4dWE8/eN7z4xpqLJ4JqYVHVWH4PCFKoGY6GnBa8ebnn5nUg+8dcnL3H7rij7qHuNX3TczweaLsKqRMuWkZ/IOdEZZZMo+If/v4eKSw//NGSR3KDyR7TJLTZ0zcjqh0Ydhapb6nlR5xyg8Hh/A1K3ZG4L/ehlEgxo39TJqaErOb2KgtIFXV1QV09QqZzhGoAi85QqYyXgehTCaHUDSkFvLtFppKg42MihExSdfHqpYSRDWlyxN4Rg8c0LV3fvVWLC8AoSfAOxdcx9W+vX4Nj069FYcoU8FsGjFEIEUnBMU6eGFRpw6RGqZqcBFlMhDgVNIcSiNwQxQXBZ6pqAQxARuoS+xyJJBHnSJxUuNfTxHLRchq2tLWgbQ3rszMWy4AINSSNlA7rTwS/ZH/XcPVV5sU3Wry7xGiPNFu+/f4ZdVrw0787cHkdEarHWouUGu8DzvqCQ9EGKRRN1bxdWY/DQFs3xWzsLN5ZQMyyA4HICi0kQzghhSClkhsmhKSqDJmIj0eCS2QCVZPJwpIJaJWpTUKJVIbJjaJra8glzipmh0i5YHQDCGUIMWNqjxQNIjacYsBnj6xaEAYhBpJVkBsE5bNyslQm48YTbVWhBLhsyfmENp6zTUbrEWRifab55nc6No88dT9iFpk8J/AGV4F4xGcvj7z50jOcesJUo2XDdvclskklftx2rBaCpjVM00DMidOwQ1flZF2JzOkYydpwfZX4d//2hu//sOejby7IcqLKGTkFRLhE1Q5rj8RqC3xQ+Or5Dl3tUNUAOpWEGKURWaJyxgcLYcbOpBYj1qwWL1DVU9K0IaGplUTWt+z3l4iwR8SRTguENMQIAwMyCwiSkCSkFqlaQkpEO6B0YbSPCHzw1H0hOtRNg6nKGELLQhXIYU4F/jW73tnCpGbl92w2QRmDqRpyCoWO6CMxKKRQpDSHK0ZX/Gd5Zg2RUboiJ0P05VTh7QGRDXUTUDISo8WFfbFZyII5IiWu3rzBPK9otUH2ms/snrrt6ddr9neOlD1SanLSHA9AFEz2nm654ezpBikvEZWFtOXFR0sQa8Yf37PfeZp6QcyKEBJKiEJvtLAfjugq03aayVmELHHbw8kWGmEuOXhaKbRqiDkzWV+0wDmBjGgVgAAz5L5qPE0jqVuFqRJS1eSsaBtNYygq5Qb6TtLURSYRUkRQbA0plofA50JvdFFgx8j2bo/1vlghREQbh9RHsihtiXMlMDPngtwIWaOBLDy6Gmh7R9cnzs5b+k5TNZnVRtEsBpreU7UBaTIJTYwronvEX/34ir//my3JthhTIWVA5Hu6Rcfm4oK73YGcM6fB4r3FB4/UqgQnyIgynuATqhXUbcPqvKNq93zx1ZH7IfDhbzxhvZqI4q7o1+xI3d4whr+jikdizGh9Tds6dFWCKJUSSFFO7ErqOaVEEH0ix0xWG7J+TpAXyHpBJSNCVfjDLds3f06YLtG5tG0xCrLoynclNFIYlFjQ1c+Z0gKyRApHCJTo9jGigifEgJ0mlssN/WKJUpmQ5mchfY08+WULC+IhIO3d7Ofe2cL0tYw+v5XRex9Ky+Yk5bcuEELPiIiHZJQCIxLCUClJ8BlvA7pakoMiWjkD54+E3KJkAixSVVRG4XzEyExbdwTrGQ8WaUoyyOmUqPuOcO9QRjHsR7L2ZFUhk2T7eiL8xRv+8D9/j8V6JOOROpHylqfvL/nDPzzni88mLi/3HI5FYazkHPcTmLGXEGOhKC4XqyIRIBJ8ueFL5DiIXDZYdWVQySJFREpP2yfaDipdUZkaaST9QtN04PwB5yakFKyXDU0lqRvQ2lPVGa3DXNjlTLxNeBcICUIuEgznMzkZnDXc3Ah2hwRao6oK7yuczYyDYxhgmpi5QprT9ghYLs4rNucN3cKzWEnqxrFYJupGUDUBU0VMUxYRMtek2HP1WnF7E/n5x4FpaOjMgqqSmCqTYyj8bdnh7Z7AiJsmsghUNfgUqDvFB996n825pms2/PQ/fIJKmUnsMK3n4sUFt3cDf/bnV/zod5/BkwlZHekMiDSi+lsKSSshzYgWA1KWjViOAaF12cLNNGj5kEKToG7XqOoxXi5BVCgcUgiOx2uSe4mOd1TJIFUkCEh5Qs4xSxFBoEeoipQXpNQC8/0SC7U1ZUHwkbptEBH8EOiXHaKvqKq2AOXyf1SUeIiTfXdPUu9sYfrVL60A1KbJIYwgRvl2LZpyKV7F6FtR0LoeYzTB59ICioK/yKnnIXBHZEPMNUIsIFdEPyGTpDbQisSL955Sac3+5DhNI+fnZ+z3++IeJ3Fzv8VOnr0bWTUraiVJqeLly4H/819f809+/ymLTUBVW1D36H7H+fOOfrPm0fWSu7uEdQJkRCrHOB4RUaOkQqk8864r7m4HbsKRyZZNU57Bc2E8IlWJMFqeRxZ9y2k8sj7LLFcCYyJ9r8g4FsuM1JYQp2KIVYK2cXRti9IJOaN61cy21nO+vUh5TlLJJXeOgJy3QG6CnFachsRhcByHHcM+cTokDvvM0EicU5REm8ROOiZ75MnzM54+TzR9ou0SdStQZsKYCiEqoCZGzTQlDlvNVy8dn356JGdYr5/TmYwdJgQjTW3o+yXb+xPH456cI33fsru/pl/WhYPVVjx5seBHf/gdzp8Ylu05H33rGX/707/mON3x/jdbPvzmhifHC/6X/+lL/vxfXvLHP1pjniVkHwg4wKKawLIroQdZTBA8utZFkDknOKdYBs8p5rIRFg3KbED1IDpy1kWhnTPT6TUyHWhNwKSMkBGZMiEntJmtJllBFvi8xxhHEi3OUza7EnIK5SUiwZ0i2WW8icToCKmjbSmImvS1v+7dPB/9P693tjA9aFcFzKGKsXjOYvHIkSRKFsSo0UXHIZXAh+IpSjkXsLyYZ5BJQq6LfytEhNCkXCFlRqsVusm0WiO8Q2fLerEhxYl+UeOyp0se0wSyPrF+UhGkJUvBaTdgkQg6QqiIUfL5F5Hj8JLv/NaK9z6EqtVlkhUCh2HHZDXbveM0RLqF5Oxccf64p1YBJSJNI5EyktPABx/UnA6Sl18dcNMJJRds7w/kHGg7xfMX53TLhNEZxILlRtB2GTnD0bJImGpCSAuysKyVyihlEdLTtoblqi3APF+i0iulUEIUFrcsp8hY4NOYBrQKuKkA750vKJHj4Nldw/Y2sG1LGENMJaW3biT5Wx23t4Hv/gA2FwGhAlJnKtPivGE4wW4v8Lbi+s3A9m5kPBqGQWHjEiEVq5VGV562beibBu8HrN1TVZrVssP6iTQTSoVQLNcLVJ14/sE5zz9a020smh0veo2vV9wfWxbnnrS8Zrk+47/97/6IT/5y4uN//EvWZMyjSDdD80g7VD0WG0n2JCFBa6q2+POUEEVDFtOcLScgd1T1OYmOmE1BQQuLIBDtjq7KVGRMLBhpIUARySqUVlpADpGUI1orEoqSoynn5yMiRZwz6iRJqDmOy+HjhHWBpl3NmN63DxbwLp+VyvXOFqbyyv5aKPawGs+p8JZKHqPmIYI654hIpQFUssgEymY8z/B9SUwUwZtsyqA4zRA0sSxZXUFAmkCMWCepq4a+z2yHG3yynF90oFuu7wdEN7C/TdSLGizEyTONiRg0SRrudomf/PSWy7vM0/ckWmmSr9neRl6/3HFzk5HScDx6jvvI02dLPnwB/SLRNImmFQQ/IMXA5kzy6JnAuRKzPY2GGAVKR9puV/ja5EI1lA5TG5TSfPDBC65ufo5UAVOL4hmUgBAoJUvEuvD4HFENVJWckRolkZaYkSJRVwVdJmQmS4uUnr7WCCzSRWSbqVeCvtdsHtfcXAb2u4APkbouxbdqFd3inMyArj3ImugNdljw5it4fWmZppZpMFy+DGhWtHVNDkeCj6wvWkyd+NZ3X+Dtkem4I0bFNAp225HLyy8IGaq2oukrqq6ibmow5WVgmkwSe3bDFa/e3DCEPWY9UK0SqoEY73n0UeDbH/6Iv/yTHcP13+FqjRYSIyP1WqF1JoUSNW9TeXkKA8oY/DSUe9IYosslsVgammZVGPJZkWJA4cjhRPQTSiS0EGhh5uSfTBYVMfuCvDeGICukXpLNimh7EhNQNtE5J1IuyxCR1UzGTIXDfpzwoczUgvd8XYrE277uXT49vbOF6SHnK+dcPG9SorUh4FC6mGQhzsJFUfg+MRYZRyqxOEYqoKxcU5gopAGH1AXRIZUpAjQHWurZTgAuWg5Hx1R7knckxGySTZhGoLuJxSNP0xtO20A4GfbXEXeYcBFCMCQHgZr0MnA6imI3GQf8JLFjQ6U6hNJIHCJFbt5ERAp88OECKRNtD00nykkoZ6oQ55X0hPdxdpOXpBEjFePkGLzn0aMOpKVpFLvjS0wT6XqN0gGlmcMcFDHHB6rr/H0ntBEoIwsSV5ZggJzzzBIqywitzDzTi+UzjABVZlJ1pag3Pe99dMH2duDu9hKtHXXjqBaKlDxKVyBahqHm9krw6vPEF586fFpwmhTPn32bzW/WRBvY3rxEaTDO8fjRgsWyQUnLGA4cT1fUtUbqCMIWD2Tds1jWYD0+JfbHkaoTfPjBbyKA437L4K4Z3ZYkPXUNMQoq2ZNoELUghnsunmpev4Jpr5G5aJHqukLVEpECWlZv/XQlXCHO5lkQMSExVNoQ0JAF1lo8E5XMkE8Ee09IE3VVXpwplFisrAr9U0sDlCDQYEviz2K5wOaGrCH4iCbNdplYRhkqgygYHJEkBIhOMqU93rs5ECL/Mmbgnb7e2cLELKzMX/9v8X9VGqECiYxU+e33nFJRyipKXlcOgZwlSsymRxFLUZKhtHsKks9IBEoockiM1qJkREtFt1xSN4Eh7jAmcrZe0PWau+OBm+GKfiMQS0Xb1ezeZOyQUYeSQOKDI06acZLEVGF0R06BHBRG1Xjh0bpFqDKb0LIof69eTZwOjotHgsfP4NFjzWKRkeprOqGQibpV88asaLkAuram6xtM7TBNRpuiXzKVROmA1AEpH+J+Ig9gPUExfyopiTmVlk0JIrm0c3NajdRfIzYj/mCaAAAgAElEQVTe2iAyoNUcV1WMxH3dIpPgUb2iWYzAHqUdNk5I2SJEzeFQ849/f+TqVcv+RnM6GFTdsDm7YLAjMg9EO9D1gg8/+C4/+/QXpDDy1RevuDhbsugqQhhRWs3LgqIoD2EkBE1WAq0bzi8e8V//N/+MD7/ZcAp/x+kwYLGMk0fXEpImjAmbDYvle5yf/wb3X9xxt72EJNnfZmSWYBJ1D6GNJDLo2cspEyUGPM8nbwFJzIJWj6wzMTicHZnSgawiKtzjpjtQjqzKtFzI0gEEmUnSF0KBkMQwv1RNQmpF03WoJLCTAxFQKRNcLFQMAikVC4pEkWMiIok2zycm3pp5v77e3TPTO1uYsixv6kyJhBZRoWOLSBNKCyIJmWOh9WXKJg5R2rX0MOQLhYkT/TxzKnFKClnSVmMokoRQLARaBYKz+GQZp5onzx/TZ0lVtYzWMPjMrT2RZGS1XLDsn5JDxZfmFseetTQc7x1KNPgg8C6w9yc4eRZNR/SZzlQolYnRsuyWxCQIXgAV3sE4Ttzdej75ZOD8HJ6/3/L0WUfblKO7VJoYxdzGJOyUSM6gK8n6TLM4i7RiQmoPMoKYCs627KSLbVPODOkoy1uXMsuQOSOLBxXxdss8hwylMhDPgRmZW05fQlJSgaVAZAnC8+TpgsP2FiETdihv8q4R5HjBzeWa/+1f/ILgLnjy6Psc7m5YrROqqhBScNjvOO4OBGcxGYb9oQQ/ZMF5V5PtCdlArSRxShAr1t1jsqzYDkeO48ApSB6/V/Gf/dPv8cH3FD5do42miY9JU0TgMZVA1xGjDYhzutVHxNzw6ssvUSj65ZLt/ZZ6NEinmVxEW0mYI8CFSshKgfAYZSDXiOyBMsQWWeFzxTidcOE1g78jZEeKe/zuJVIcQAhErYrzIILMJWcQzK/MU1eLDV23IIemMLu1QgwT3kakEej5NBSCAAVKzwJw59BZ09pEHR9e3mVKQhYkUTLt3sXr3S1M81ypPBYlAllikDLOW4mSZsq8aUipJOrGWLhNRVoQZtB/mvU4gTiv22MISCDFUNJkoyf4EzlFKi24vnxD21iqZkvbrHjy3jMsik8v/4H1Wcf6rEVkw5s3d9zuXhNVZvlI8ez5GRLF5eWuREI5gZ8s26MHV2MFVDM4P5FLbpvI5RQiKpTQiCywQ8PV5Ll+PbFejdStwDtPjJmcBcHDOCZikCASUkaWa8HmQvHBRxXf+NaG9Rmo9T1oSZZlziCkQuTEA7wemR+E9HMmHeSYiSmTJbNL/gFfXDCvOaWvNWYhvfXf5VjmH69f/mImOp5ARpQypHjO4W7Nv/nTr3h+9juktGK/D9hhKum7KSKEpK8rUmU4TuPbe2A4nUhBkdLI5mwBMSGlwvuAEJK677jdnZhSxOXI4/c3/Bf/9Ae8/y3NyZUZ2/n5gpA79vsKO0n6dcdmvSCEwKJ5jJQLbt/s8NOWF+8vGULm/uU9pykgRkETNKeTRTYJpQurSwlVRKQz9zzFhKIk5yINGcdwumbIgtErcEcO9h453qF1gDlUIIs8/72X8EqpCiVThHI61UpRNx3eNUgyIWSatsfohB0dgTAvhxJCaoK3GCWRORF9gFDCXEsH8vWs6V1u6t7pwlTewGVOJIWaDY+eHEt+vBQl0E8IQSbykDidU3FZe19avSKEDDO3ORdP3EyD8JMl2bEMz7NHS0GMltZEtrevMPINul0whpGdjyR5ZLWp0Y3k5Zev+PTL12St6TaaD99/gsJjpwnRw3AUiLzEDZmrr/aMY8CHES89IUZGO2KMxjlLXVcILKSM0R1K9GhpSDlye2tpmg7nHYfDHq00RjeIWcsVswPhYR8ZjwP3VyOf/8zxnd8857u/u6ZbTrS9RQgHWhRMhohlrsHD1rJIL9KcOffWnf42977whZg7OCXFvFgos0CFQIuETMUoHcM8+xAVgnNOdy/4sz/5OXL6gCBWDIPl6s0l42HLJBJt1/He++9zdXkJYaKvNd4X5LCzlrpu8N4TvEeJhNQVOYI0kr3fEmpP1pEYHHp9R26/5OXdxywcLPsFYrfBuvI5v/PDH5F1wMUJjcK7JbLZcPXpJ2yWmVbuiV3GJ8F+ikgrWQdZBJBRIAKQBM4GVE5oJWfmdiJJyqwHgIng7vAhFfLmdM/pcEknHedr0EahTZ63aml+AZTljpACXUnyyeGCRRtNTUsKkWVfl/RkGRFZ46QnugJADCGDkuQYS/5cNKS3Y5Ff5TC9u43cO1yYytdWRIV5FhYKoQCJUoVRo5SYU0gj3ru3ehJ4yH9/aPNm7eK8wdOyFIPgHcHawvYWGaXy24RZTSROB6bxK7ySNH6PWK1xYYtNHps0IaZSDEXk/RdPOH9kmIY9ZhmQnWZ3L3EnxTiNJJmQjSaPkYTEJUeOCesFMQVc9MCJttbkEPE2Y0xLFpLDcGQIma7ryNogm4qYyvyCLBGqR6tETA6JLuzxG8e//bMrdscV/8mPzrh46lDVHpOnMm9SxdbyEMUu5Ny0PbTCoojJg/zVOUicf8xJzFFARbmOAJUMWQQioXwvuSX7Z9zdL/kX//wTTtuO9x6/z7hLDKcj9nQPfgCp8GPmq89+XgSZyROsm19EGaUkh/0ebTKn4cQ4ZWhakhIkPdCdSf7J7/0WQST+4eN/5Mn7kOovkW1kyp6z9gVN1xWtViUIacDZjJRr7FDRts+4/vzIT378V/zxHzhEO5JkRbVsGa0jSPABjFDEFKhUWRDEWCLVxVyGtCmZby5kklTlfouRbA+kYcAerwjHO84ftcQYMCKS04MouHzvSIEgIcuoD2lKZJMxmihqhC4ZdxJNVAWgJ6QmCEqaTi5zr7dBT6kESpRc36+Flu86Ku6dLUwPAX0Px00pJZeXl8h6YnOxKaCzWE5GKUUe4sPLgLYMaUsooCKEcgyIseh4ok8EF4lTKCvxXB48EmRVNoAxRZw7obJDKYmLO65eXXO73TG4jNYNp1NGa1itO9pFZnBbUnZ4b1F1g6oE96+3HMdAtzRQN/h9QmNKQc2K43Gci4tE5J5pAk0ZKFvniQKmkJHZE8VIFpICSIzznz0hpSPkWLZeUiBkwzREJmv4q3+34+bG8nu//4jnH26o2i11G0u7kBWICLJkjj1sG2Iuzvgs5nnEfAeneRaOKD9HCcHDPzFCihVSK5AOMER/wfbynP/9Tz7l9WcDMhmm/eeQI/1CsFyULZKUBqUNWc4BoLl4I0UWVLqYUEv0u+ZoB5QuRu3+rGL9VLN5rlg/3dOvGl58+5tEOVI1I5n7WTt2i3UjOWZCdgjZsl58gEpr5NCSjpqf/MW/5gffXbPqb0nBUS8X/Pbvf4ef/eIrvL/j5BytlJhGYDqNrhPKZCIZYqAyFeREDJEYS/EQFBV9GCxMljgMGJFRQiJEnI3SlFZcFU1eUZSrcnRNmbZTBJ3QVUaLcl/kDFQCRKAWhVcviEilySqQbFlu1J0hDOrtEin/UvP2LhcleIcL00P7kLOYC02m6xpETcGRCoGU6uswP6lLmsR8JE4pfY13QBFjIvhcHP0+E10qrUBWqFRkArKSGFmVt44oDwdEtNE4GRjcERcC1oKfPNFH1hvN42c1Qg4zLRCM6Zms4uZ6z+FQtoNaF6Z3Z2oqUVNXLXe3ew7HE9rUnF885u5mS4wgkFTKkEIkhEhIhUHth4nGaIRPGFWigUROhZBAxKiMCxkxKWKo8FEwbBWffxI57t/w2z98xLe/e8Fio6lqh5KUzZb0Mx6jtHml4Z39FQgSDw9TOUbJXH4sf0O8LST83+y9Wa8lZ3am96xviog9nSkzySRZpIpVGi3JbaNlNGDY/sW+sO/chtFo3wi2IAiGG22pVSOrilNmnmnvHcM3+mLFOWQJvjJkIwkobkjmyHN2xIr1rfW+zyuWkjrE3pDill/9vPDX/+s/8Hgb2PevGM8VSNhQSCWt276gFMeVgV2aRhVV0Uy2vlNtkPgNblC8sgkG8RE7FF796JLi7jidj/SbF2z6LYtUbq4umWNjiTPGBKVLSqZJwRo4n29Jj7CRj/ji5z/nwxeFjz7OCIkUDbFlrl5t+WzzEx7fBjxvaSHhN5bNIeC6RpUZu26DnTJ6aAKds9SkFNOSQYpgStVBc9Ga46ywplnhPOvG1SDrEsEHi62NgxnIrqe2hUZRY+4aIf40+wu9Q6RDUtIclRYpUkmpUDXO5rk4ve8F6el6fwuTPN32q+jPaC5bCxWxRTum2jBiVz+Qvg2e4mxaq7AaZVPOLLNCs2pp65HPqx8sJr1xrKMVoWQIISB4xDpEHJnGuEwsOWGsxRnLMlZ813j1wQ7fLcQ0UZulFE8rhvvbyOlYSAt0LrANPQHPob/AtsAyZ5wXDhc7WvW8ePGScbzHGcGjqvauepgjlDUYE6G1hXGO7DY7RLyaVLEYLN7ZlVUu1OzJebXyTJUvfz1y/+4bvv7djj//VzuubgybXcSHsg7PNRJb6QzocHtNCQHWI3NjBUFp8VolHIWmP+wXatox3V/zd39zy9/+zTfk6On8hlQNKUVSecTEiO8s3m0o1TPNCckLrguUahXv4nq2vXKNUs7MdSZ0lv4w0G87PvyoA/9At4n0+w7rGqe7I+FyA0EwLbBMQuguQBRdHHqLF6dzyVi4ffyWn//y10gs/NVfvSaXX6sbP3tOc2JoI9U7WjDU3LBecJ0gvrJKjaAJTiw1Zmxr9N4q1sZZlrrGa9VMigutlOf47xURoRs0s6Ju7coWq2b13qmspZmCsSrfqEWxN6UUvLfPw/OGpgK5ZjT4gAadUKr7nuzmh1KW3ufCBCBtHYCrFOAZxN50K1OLvgOesttL0VkRsHZUTxs9g7VufUtXjKtYqhYqI0BHroa7d3eM0z0f/WhPv4+w7Sh+wzmPTFiqNTST6beOfuPZXg70gzBNJ3JsUD2tbRgnx+27hWlqbDcdg3TsfccHl5cqfMvgMXC9I9iZh/uJN99+AVKJuZGbqCnXKpTFucphu8OIMC9nYpfwQ8SHyjQVWjXkYlkWh6mBsujioBRDRaAkWhXOb+HtQ+Xvf33mP/vLDX/2pxs2wxHvIp1tdK7iZAO+YPqKgsqKziNWIF82lVIFyV5NxSSKCBVHmrbcv9ny1//+S/7xP0bi9AE+bFjmzOITOedn/pWbLN5bchFSy5iWOZ8KyyzEqHQB5ytSLY3Axcstf/SXB4aLRMyFjz+/4PbhlrndExC6sGXYXOCGLR0De3dNcmBspN8YlqwYZSFRcsLYiRefOKSN3HQX0N4geWJOnuMJYuvYNQ3HLG1GWsbbijea/9f1HdKEVBdoGgSBNDX0RsBaJA5Y0yvK2ASs7/AkgldNmXdWN5LkFXXDqiVLzHMhFcuUK9IXjE14U6h4mhGSGDQiOSrrvDWNGKPiLOToiVRys5T17fJsDjZCFvD/jw/d+3G934Xpe9dT2q6xQhMhZzUw6pveI2KgJUBh/CKy4mS1UDnrkCDkEmnSKDUhttKMIkgrwsXllv2lxYaJIhPYheqEucDSCuItJhVqSdxc39BvPMtyZui3VO84PQhffnni+FB5fEhsh45dt6WeE9cXG2xL5JTpwwZyIvnIJ6/3dK7weLzn5nLL6TiyTDPdcMGLly/5+m3iPI2k+sDV1SWX3QXGD7hQGDZbvv76ltt3ZxqWZVEGOcXjTGCOekQrrQGOUnWm9PXbkYf/7cjbt1t++vkFlxeBTT/SdfrQ9LtGMBXrBIPBGi1KrYJkjYSqdaE2iNlQ24GSr/n1r+B//+sv+OZ3lbTsiHlmjGf6Xh+meZnJRT9D6xquJZDMH/zha3YHw7zMeD8QU+Y8PSBG+Oq39wTn+aM/+4yPPjKMy+/YDoarPrH3e1ICSqUdHQszzI9YDLcc2ewPUM9witS5cPv2EUOH7w2X1xbvMy8/v+B8e8LROGxu+PbNxHyOXL/8WBOP54llnoA1FklE78GmR7i+65+pk3WNvRVWDE/N1Jr0pVkzJWdCQPVkRp4H0LWxZtfpTJRqaFk4HTMPY+THf/oJ1AuEfg2oABGDtYKIf54dKY9c8/VaLYTgsNXrQmhVh/zLVu6f5VrRJ6u+o7VCKQnrrc44WnsebINGHxnjyFkLUs5ZmUc2ULLqZJ4zw4yQpYDT/LcSC30XtJ0OGbMpZLNwJHOfF3COanVLcnGx4aOPLxlPj3R+Q0xCXDxf/u5rvv16phXHvt/ROU+dF0zJPN6+Zapwsdux6SoWDaLsOuHw+ZZSNwz9jof7E48PI8Fvub7Zs7sQfvXFb7g73iOPCx9/+jEXV3vuHr/G9h03H+6IpfBwPxG2gdZ1jMfMeRkRa9TWUMEahwRLtwtUl3Gbjr//5ch/+vmX/NFPLnn5yrI7jFy+KFxa2IplEEMQ3RgZHKUoj7pWPXbEYoh5z/Hhhr/9m3t+9g+RuGyoxWCDRdrC1X7DZtuxpInh4sA8Z5pUdvuOfoB+V/k3/81H7K8am31jsx0Yp4VlecV59PzP/9N/4MXhQ8Im89XvvuLmujHenTA3DVcnxoeJlCz7y0skCF9/8SuWyXB5sePqcuDh7h3OWAyZ8ZwZHxw3Hxr8x46XH6hNpoxnTjPU/kBdHME6JDeW+UyNibwkvNPjbC2Q4hqGueq6dK65FplSdVvXQGyhpIVSIrLqm4ZBASpPYb+1VmTdLj+nTzegeXI0tLrn8vAZcEVuA02SBnI21UyJgF9norrVa5iiguEihowj+PD0JOlT1f5lK/f/+tLv9bODF9YFKFLJSdNXQ/Br0dI3WGuQk6q7S2nK9FmN1TooV+VryRmzCtlyrqSa1jmWUcKjXXgc33C/fMkcEk0aUnWYuN32fPzxC2gTnYeGQ+j53Re33L2d6d1A1+2oYyKeztha8E1Vvd52zONJo6s3gavQ6Ad1j/sQIB/ZDsLrDw4ao2QmxDWmOeD8ns32QCuN8zgyxYW+JGzfsbvY0MRwOmZKrry8uianxjTP1JIZT4lUElXO/OGffM6f/MVfEtOR3/7qC9589Y4vb0989qc/5vr1Dd3ujmJGZgrkQm0aa6VvWw8tUOtAnHfcPRh+/otHfvnLr6j5gouLD5jnM7lM5DpzuN7w8uVLjG08ThMG2KMI2JtXe7ptw4WF33zzM3h7z6efb/jR5TVdELrqGb/1/Ov/+id8/OJT2jzz5ncTeX7Hq/3A6fbIZghM943WOsz+wIvrjs1honOaeBPHBTyc7k50YcM+XDJtC+V0B0co3uj2b2zUSbi9O2l4aPOcyiMpZ1JOmjPYKtZ4tluPtVVX9q1Qm2Ct4mpyyWhn+bQRjtRWKDVTW10tIW01Suu4odb6XSx5080xtWJNwInHmR20YSVhdFRRRbgxdq2EFdaiZ5zFGzCl4awF71kQFe+ydnLvtazyu+u9LUyg26nK2vaahrEgVkjpCUC2GhPRrumpOW5NMGKoooiPFBViX4ruWr11ut0xHuv0iKIjXIBGzJHiC9FkYsl477HGsdlv+Mnnn1LLwjIdCdZwHjVM8d3Xd2y8ZryZtBCniU4qmw5sA0Oh5gnfG4a9BxlxtiA+sd9ugYhFh/MGD7Vpy24ru+1MPAuOSpzP3J8eiHbGLYL1gf2Lgd3Nlru7MzE1jDmRS2FjoOTCfuxZxsa8RLpN5C//i58Q0wOffrbl699+zc9/9gt+9+Y3fPCT13z0B58w9Jnx8ZY0PnA+T6qpMULDEWPl+Gj54peF//B/fMXxsfDpjz5ld3ngl7/+JSlP3NzsaA22m7AGanrCruPNt3fEqINyuX0gv1k09dh7PvrRhponXr3s6AZDivD2m6/49LM/4no3Mz88kC4aXz2cqFPCe8+2H+hCz5uvE4933xC6V3z80xcE/wYvMHeVC9czbwLx3KgpYrsZ3wl7K3CqSLGYxZFPiZQqY5zIEhGbQES3vgLNC1YcrWRqqTTTniUtOWYMZtWA6YuzNWhEanu6T8t32ifR6C4RozFcrWKs/ndr2r3nWFhmcGFQH6C1qkUSSzMq5zCwmtsdhYJUsFhsMdRcqM3SdQGzZtwZnncXvycDeR+v97gwfddsNqCUjNL7IIT+mTHzNEsCaFXjcHLWN5ducFeIlxg8DqRRYqEWFRhWcat4sKwObKUn+n4gmGGVQwuuGT64fkWaIjGdFQLfPEY8D3ePUBqH3tKmiG9w9dGWzlWkKbEwFwP0uMGBizQKuU5Y58hGMOJZ4oi3T3RO7cS2m44ff3rgg4PhZ7/4HdOy8PrHl2xfXfB2vMX3lcMm8/qjjzieINfCHCfEQcyJh7vIeGs5fmtpD/CLX/yC/+G//7fE+MDnP/6QP/vTH/PTn77k27e/4Pb2Leel8Or1jk8+2mB2huMtvHuzME4zc0x8e1f46ndHvvoNzIvBes9pvKcf4MUrg/d7jIFlzjRO0ASRjmBh21sMiXnOTI+Nac4423O1vUKi8M0Xb/nm0yOffX7NV7/5LXk+cn74ObYYSAu7g+FwMOTZcHUYeHG14Wp74GACv/7ViX/827d8eP0x7qXgAyx1xpqKyMx250ix0GHpQ4eTRotqrG6LUCa11lhUkpDzQlnNuUpaYN1aJkouuM5gbVtX9k9Xez7SCZDLRExCqRrVpV29Ll30ljV65Hp2/KsB2FRHnC05Brp+Ty2OalRHJtau44hGq0WP2dYg4ii5UVcAmbEG13XEbNewhO+eq/e4Hj1f73lh0rIu8CwZEKPbN2jP6SKgxSfXpzhpvvc20Pb1KeNLo5st1ng19daCiCXlhBjBeEtKlXlZsN7gsmVZMtvOs4wjy6iwfmsUmTHNjePjiU3nyaeEbRo9fREWug52u8Cw6cEOiN3itztO8czt3dc0cRgLXejIuVFiXr/qgjRdMQe3XyPNj+y3kdefXrP/eEs7ZK6GK2zv6GWm645cfziAbyylMi4nGnD31vH2N408FWg9cTR8+eu3xGXmmy/u+Nl//C3/6r/8EX/y559TzRUP56+4u3/H648svhMuXh7otpW7+8rjGY524toK+5cHTreFt7+74+LScHHp2bOHVnT9XYMOYH0FJpZzovM6n7nYbbl7t0AM9JueYBODd/z4jz/l5U3g7t2vqOWeF1c9vTvjTKTJjHUdn/7Bgbs3Z9L5TIme47vMb/4x8/ZL2F12fP2fTpzOR16/PtBSo5aF0BU6B/0g1BrIUZNJyIa8VKZzphWwwWJbxTu1dDyp4dXyrLqunBOuru3Gs1BO1n9YnlA8tVY9BuKfuyV5GnobWX9MFwzWWRqWlIS4QDwavv16JteO3YsDLmypttMNZc20uko7aLRqEfQ0QavPTDIxlhZVdmLM0wt+nYD/AE5z729hapZmEtICNK/JKG1CqtCIOh9s2v6q2Ewji2hOIWe1UgWeips1ylNqxq4AfVXLWqkUSWAyGLduSxqtJnKcWLIGAohvnJcHaq5QLVYc5MrjY6RmwdQGc8JFiy9CcAvBOHauZ38Qml/otgMTkVIDF/76eYNz2B/0zVwrd7e3NHWeEafEMs7EPBH6ytCBtJl0jpiw0G8F5wzMiXfvHnDdlk8+v8EPsNnsOJ4WjBxxPhN2hpyVCmCx+BCYz8Ljg+Hf/7uf8Q//+Av+q//2wOtPI3sg5hO+N/gt+CFzsTcMybLLjseHyvlhpC4D0/0Vh+4FZXKkZVrJDg3vHN51pGUhp4Lvlfx4sTjy2HF+19SAWivdZcef/PQFH3zSCObMNI/c7CzNLHRdZe89rQWaVISZDz7qSSfH/LDw5jcnQu6wccSmnnz0lDu4y48YuzBsBAmWklXYGJczThw5N1r2HE+JebH4PiBB74vOB0xdSHFS0WwFb4XaCnqgVeKJEk60A8pZdUYiq1xeKqHb4XyH1EYc1YIkxmGcR9qk4Din3VPJnmXe8nCfeXh74nwPS1345M8uqb4jZi2ARkSN2ABiMU6Pl6asIydTqLIgrVCNUjbXoN7vTZje/8r0/ham711PCleFlql83xhDLfLcNbWmbxHBqRGysSpkdeORkqrFS139bQjBBHI2lAjOC6VOVDmR7EQ2mWJUfdx1DmuFZYrMU6QkzV+DjsdzxBnHPI10CK5UXDUqkkswnRc2lxbfW8REpBq23Y7Hu8gnP/oMxGJdWNXAPf32A5BCXGbSNFKXmfvbd2S74du3Z2iFQ99z2Hc4u+DEkrs9YifevjnxZZm5fOG5uD5wPBbKA7BUDoNws+3IqbGMmVIsx7FQywM3Nzs2h8jv3n5JdIaXrwbiAl3fEJuJ+UyzDmcsW2fZhA31YkOeLOfdTGgJsnCOqnnyNiBN1JO4wDwtDKmRouXhNkIyHA57zo8PnI8jb79O3L013LzuePlBz9DdAOp9FCp5GjEtq1ZHoLQFZwym97x4ueG+VJp4DjcNCd9SS4ImbDcebCGlQloLUy2NKTYkW3JOzDHiOsdmLyy14kVDI50VaIZYMkZgf9hpqGpp2Nzwbh0+r2t/g0Bt5FyJS4PmsWg8eksJu2JxW8q0ogWjlqKZcVVYFsPbb8+cj5U0w2Y4MB0NXXdJrTrjRDSz70k881xm2oquEWiiAR0WgzRLK3Zd6nxP+f0DOMv9IAoT8Fzkn95ItRRKFazxa2Faf1nTuIHn1Fm007ZOQwk1FNBQsr59Ui340LO0xDTfE5dvMd3IIhOxZYwRhiHQWHPnaMSodoxUNaXCpMLNrudiG5i/HbGpQWnkuXEsI9224SuEOtPvbqhV+OSDC7a9YENHKoD1GDsw7PcscaSRkGI4nRfysnD/buT4WNiQuLi44KMPtoiZ6HpPFM/lNvLRBzNTegTJlNPEzlpeX17w6YsNIoZpnHjz7T1ptjTxvKRnWWbEHrm47Nju9uRY+e2vRi4vPNbsuXABI4lKAim00rHf7sjO0zoDdaQ3BUoirMeZZVw47C6R5jgfG8PO0OpCqxf0XeKrSbi7vcW7jrAx9D7xcPsly7hjPneIyc5xxm0AACAASURBVLSSkGooqeCtatupTY/ZU2JwA95ZLi87lqN2d5/94Ra/ORO8wdgGtiq6JXjNZMuVmho56Uo9xsz24Bl2AduJwv6cWmNqTohZj3VN2Gw8YhdC53Ben/KaK030z6OoqblWKEn/7FIzrVriFKkxQ1UwYUuJ1hmsDUhrUC3jY+F4nzg+FGWP7Tz7/Q2Hi49BNjpDqvV5u/dsu1pv/CpNZ1BmHaqv5Ff1HZnvTp78EPql97owfTcQfH4/NKUD1JrX4aHaF4wxWKt+uJIbMcXvcttXZbjmz+m0oInBiGJTfAuKQbGR3E7M7QHiRPJRt3qAD54URw03wFBpLIs6tk1uuJzZdx271tjsAmVMeCPkDHGCNDZwEbGZViqIw/hANAvVOEw3EIYN1qntxNrEpq8cT4XxbuL0LtFLz2efvGDz4kB/dcni1UZjw0CKI80VhIXdJmh2nGls+oFPX3cYsTyeHjkdIx54881E5x0XlxeI3TNOIwiksZBPliJ7vno48/DuLX/4x9fcfPgCK5OmCJsNpXaEbsc4ntlc7JAMwQY6DMsUMYjOPErl6upATollmfEmkEYhpzPTOOJN4frS8eGH8NM/2nN1KFAeNXU4J92YRXg8R9KscouWHNPYMHtLjhNxKly/3LK/uMRvRkynnK1WKyUVStPpUEr6QvE4WqssuYAVhkOP+EKUiHQ6NjAVjEQwOhIopdENgvOi2rDWkLwWASfYdfXfWlGZQBVa1k7+dHxkGVXVbRpqXyl5TWDWzno8J6ZzYdtvmY8zJRtSrPj9QNdfgWxUKCyqkbLrRlpE1uADVLRpdQjbUMIGRqhSFVWjM/HvxTi939d7XJi+u1Qdq4WqFOUIIRbvPDk31HaqalhjG7ZqMu+yRFhplQrgshSqiitn3Wjkqim11hqc70hH1DeXA00Ev1E87zkmRPSm1FmUIc0NGyuuZAbrkZTwwUCuCIaShBwhncEHQ7OFOB5Vf+Ut1VRFdySHrRvy0rDekuZKmR3z/cz53T2nNyPBzFx8+II/+Isf073a4LbCOM5UCdRccaHQ8kxrBe87RBoXFzsslem84CXR2cSLyx5fAy0ZBplpTdDcOqHNjbdvH5kQ+r3QO8ubL4/0ncH1hmYszjhaabhgGfqBtKixtMQEJeMNdJ2j1Yh1HsOMuEIfvM5gbiwffpC4/9RyfbHl5QvHn/zxgcPLiu3ONAy5JHLL1Jyw1dBZjw2B6QjTsZAXoZ5mTRxxld3Bg2O1jqiCOpenQADFD9dUkepIyZJypVCx3mB6ixt6pnFG+gNCoCVNq4lToppGFwwuNLCNim51LUIICvWz1uJ8oOXMOY3kRTlJranL/zyO9MHjhHVjrEEaJUNcNIyiZkuOmaHrOM2R0+nIqxcd3u+IzSNWxaDGmOfq8oTzLaWo4NJawNKaniCeTgpPRelJv/lDuH4QhQm++4bqrKmsnZOidb9rb3Xz5pyjVn1gVMO06lFWEaaIurhLfUKlCTUbvBzYdZ9gbCYx0UzG9RoSWZ5QKtVhTMcyL7iiM6VDH7BNnekZwQRdOzNBy6Ihm/MaACC6feu3jvHtLaYz4AWJgbC1mOJoY8aWgVAzgcT1zvPJT15x8+nH9K8HllARZ9gPW2wZ2OwHsHdU6UhzxJiO8/mReAZjLecj1Nyz7RxjGgkmkupMXsCYgBQY/IaaFg57x+UQ2F55+j6z3VaO928JGxh2HdZUTMnYYDXpI094G3ACS1mgVpzzhM5R8hlpKKddZdH0XeX6OvMX//mWPhQ+eOl48XKhysJ8njmPEVrDGcXR0ISh22D6DXE8ktJELUKKju3esr80uG4ilhFjhGC6dVul/rOSGq1ZSqzkCEYGuk0glQm/sSRpNDzVOUzYIwQKEyVPiHeUpeB7A6u4F7MmFFfVG1lx2s9X0KBQ/flW16CIDnb7jhx1Om2dGm/jol376VSYR8HQE6fC6ThixGAsDNsBrFfPo1S975+oAuuL+lnPl59YTNqxqZlIi6aCFJ+enx/G9R4Xpt+v7So+UwStyjdW+sBqmWD9EEAoRcHs1jiaW1lBtT7/qSkVpfo1BWnVKjSxQCCYAyUnOjfgfCObiSkVaKt+BMcyFuJUccZhqnAxbOnrTIkJ5xQTIqVhjWAa1KUh2YHJ6pvKwpQWTGiKGWlQskZGZe8QHKfHO0o0vHoZ6D7asPm4p7sqNHPEi2Xb72imYUoitwFjK006Wq8CzW13CVJwoSPHRx4eHpjjGbE9vi9cXFyqu90GUiycziMv9j0vCNjBEjaW0kaGLauGp7DrerxfzdHtiPWVVCM5TlAFazLOgmGEZmgl0WiUZnVzub7sfRh59apnGwJXl45pfGSJM8dzomTY7zVOK5Wos7F5IuaJYbfF2h3j48jWdAxbg3QjRZTpXhcVNhqryv8nZrbBaAdTC8Nhw8uXF9w9fs05n6gGjjN0m1eI3+q4oFW8bEhjxPiE80KtkbQkalPMjrMO01RuEFOCMinadlWCp1xINSPGEwLUVNZOx6/HvYHzGDkdhTg5jECtViPMDXS7PbvLa4yxygp7Bh+q1eqpQH03Z3rSdoNYBde0tiJ6n58l+f6E5L2+fgCFadUzCSscbu145Al1UtGYcPNs2NUMeSFTqFWLQUpp9To1be3zd+mkKStJsLLQJKqIrhRyTFAiiJAK1GIYj5nH20SQgbLAfntBZztsaUCmouGc3niKaxRvlFFdPVI1181br5jYXJ+TNZoIMTZSWag5M8+FWuHq2nG4AA6FLA/U6UjfNuBmpvkMAsG/QEwgLolaBR8cYdittgVVmhe/YV4mvGn0+xOmFaRVSkwcz+8obWK/NWx2egw0Dqrp6LceMRknha0NxDzSJCKSyCmSlmV9ETSsZYXHNUrW/3bGUsuC95ZS9Qj40ac9811lujtz982yHj0apggpFo75hDEW1xmMF/zQMx1P2A52IVDrTHALxWgsvJq5DTVXxnNZ6ZxgxOCdIyXW72fFdwvGV1xvYapkI3SbD+iHHynCpZ0xdVzjukCsugdKTti8znVEE56zU/FiqRrfpDNmA8bQdbpsOZ90Q9iHjhwTaemw4pjmifMxM4+WVgPBOXJZCJ1XDjwd+B15Ddgs0mmGXEvEGBmGAY2P0i7KWKfLnSY0qWR0HivVqLZp7aKeOOrve+f0HhemJ1DZ97+FeiQTa1ajpCbqrvqx57ZVxJBifhaWAThnn829ikQxxLgggKuGWg2lCWISp/iGKCOhd5g80qRirCelxnlKpCwE77HNcti/oKSJKgET6vrvoiGSIeC6Ql6DEkwGMZVaI5gV91ss5zmqb09080cD2wzeGeIYuU8Jm44M+w2pVpKzlP1AlUKuidhucb6nNEsY9jRJLMVgxSsMzxkuLgP+XBgfHnDGQJlVhb0k+rAl+B7rLDY4cj6v9okE88J2FzBGmOPMspxxVmjSqCUhrdJ5JUpaY5R8WVevVqsaN45DmmFJhZQLxgWkeY6PR9IU2XcazCCtEZyw3fZ0fWAsM3bwhG4g1kJBkSDSZVJVo2pFN6A0gQK2WUUut0bNjTlGplmpmP0QqC0yLg9MZWQBnN/Q71/i3IXSBOKCEYsz4KSqK8BAyY2S1umx0QVKTlWPd60Seqe/xzic6VnmTDoJnes5jSOWwDdfV76VkdevD2w3Pecj3L6J9J3l5maLUigz3aajOEfYbFC8uKxz1YY1lr7vfw8WJ6t+Sp60nmJUbKnc5987yv3L8Puf9RJqaetwW7ditRbFcTxrnKryffLKEFpZu08fxFN3ZQRSyc8d17JEgjhSbjRrEN9odiS2B2rn6CkaRV4r51OkFUctjVgbL3Y3BHegFcMcCx7FqGArS6oKzQ+W8bjoUDcLzhtSzhriaR3TWIlJCydu/bqswWA0rSSDFKjzTPp2IrcKDsbBIr5hrGG38WRjqcaQtltK34P3pFYpeSH0PZIcgzGIPyOl4WzH+a5x+8Utt3czroPuAK8/29FvNdzBUZDWWKaFpRWsheA02NIYTZ9pgEhV71dTVb4V3VS1CkZU21SLIKbQLJQaccPAy08OjO9GLjdbvPfElIg5stttaBRMccTawFX6XQe14azBuEjNEGPSPLzUiGPGJJQkmXRjlotm55XS8MHjvCflynGeWDDQXzNcfo4bPqSVHciEyBnTLDUXjQdDBaOawiOaRehkZb0L4iqbbU8IYEQTHVI8aaGtFqmW66sNp2Pl9cfXfPXVPXf3C844tsMFb8odt+9OQOLVhx2lGeYsdNsNh+sXJAxtvc/NGrZR65MhmOeXLE8/Vxp1XfA4G9CfeEYW8N6f4dbrvS1MbQVnyff1F03RDqXl51/3xMYxIuvWQj1PtWgRck4U3F+KPkylPM+gnjqouS4UaRgfMJ1BTCbHmXmq7Pc7vPHcPj6Qp4qkQIuZVAv7l5cYGTQCKUCMR43hDgZkoSah2YY4HbK3KpQiIBapQq3Kss65URyEsEYlGZVI6Aan0TqHnfSN6LzT7dBj0b/HNhY3YYNBOgNLpoRHmmsK1SvCYpUp7qxTFGwTWtpx98XMUD03Q2Vzadi9AJFHatbopuAt1gjOiFonvMVSMKZgRHVCztl14aBQP2+1iFNVJa0aRKGZirFaOFLKFBK5NuwgjHWEqC8ds4Ezj3ivvrDr3TX9Zsf9w4mH+0dybbppMw0bPB++esXp/sxDeqSlRl400VZ3/obSCsYJrnc0a7l7XNiFPZevP2ZsO/zmE7AXK+Oo0XvDHBNWEs5WsmnK/HJ6L9lgsN6Qi7KWnAAm66qerAhlZxhrZLMVJpOAqjHiOXNTA0Y8qZzYbzbcvOjIyXO48Bg3o9wnozSD0CHGr8bbojPO73VJT5d2RDozFWPVR7c+L9Zqkg7rmOqHUZbe48L0T/cHxhiMGJqsSbDPqBMVtWG0ODWpa3clKiwr67ZCVn52roqiKOs8yhqWPK+qcQuxYJdMF5UIuD30jKeF5SEhycJccVnY+I7r/TUikK1m1alHOKo+xRstlL4iDpY40wWnc64VMlZrI6e6ivIKJAjB4YNyx8uam4YY8rQeQUvFB4NZZwcNqM6Qxwq2MD9OdFtHtZV5brTcUVmwHsSDcZbQBabxjjktbDeWi5eG4dAIu0J1FTs4WmsEZ7S4GKhkRHTr8/QgWCvP8yHVmOnXJCtmQ+X337HDrYVmhOAGinhOyyNSHHOCaZ6w1tBvHPvLQCMRug7rDMfzO+ZlJOUJUwItO6xTE21tQugC3jtqqaSU1khz7ayxhjAEpLMUI5Q0kO2BU94T9p/ihxf6NdRHTLmn5VusvUPCkeAybWu5vuzwtlBZMwBDw4vq28QLxjSsK3j3tCGDYeugN3SbRsqJkhvz3Mi1kJZMb4WU7zhceVKswJmnjj6XRBc8dd2swVMHWvins6Gnjqm0rEPuJ8Q0OiqQ8s//ZP7/cb3HhenpaqvrX0FdJRdcWKOq7YpzEG2jxYDUgg+OFPNKCmhI1VCBJ0CXcw5aXUMOhGB1Ne2qQ5KwnQ02BlrrsYuQTpFQDZJ17eyK8OrVNZthQ0oJOmEpBeugJoPIGkGOmjRDb2mmkatqoXLRY49CxXR70mpFcoPcaGnRQWx5+pqE1hus006y2ELwuoHMuTBH3TKKCOXUCOcnR5cQcwaT6baNPuhmsDnYvfJsP3Q4X7F20Zggq4wr5wRqhhKf2d7GAFLJrZHXIxUrpsP7oG/yqrMdu840RJ4EsTqkFmOfj6gWg8OxjI0yOcZHRc+UPfhq8R0kt3AeF6xXvLC0wjRmWrR0QwPJ3N3eE6zDONWF+c4hRSkL4i1XNxf0F1veHe/BGLb+grC9YLh4jR1eruEHj4j7FpPfkMsbNtt7glWErhiDa02tJLS1Y9K5XegMdRVWGrMuDHLTlBkB64XOG3ytpFgJQbd5t28nSmk4V8l1xnUWimCMp+s6Eond4RLvO9qqjqwlgehp4PvIn6eRxHdiS5XB8MR4+sH0SL9/vbeF6YmupDe42lCsFVywVDNjpJFLxBiLMUF/ZdO4nFYtIpYGGCdQF4wXOgtiCiJ19dBVShGsbJCaafFEVxsmB/ooOJOp7wqXGRyG5BpH24je8vknH2NtjyFglxljM1jDXBLSDZTcaB3kPOM2Hb4VTM7azVVHThrzY7CrSC6tc5NGsW3VWjWMrYhVUJlzVmcX3pCszniKgVi0U3FefV5La3jfsB6uD55+O5AY8VvB9bJ2cTPGNRB9qBB1vSNayM3qu9LvK7h1keCsFhppgqmsAs1Gy0pG8GukVKsNMKRSV8SMwwR40tnUaFiOlXyGeJqxFbx11AgPd5Fha/FboUjGpELXeSyVm8tLzo+F23f3XFzu8P0GWtU8O0lYb5hTBit0Q8AMHXODcxK2+wPeX7E9vGIz7Ck2UsojMf4G0pdc9CfoJmiJ4HSDmrPqqrzXYmsdmDVAoBRVh9eq3TpVw1lr1dmfffp+FuVZlawSkloM81mPucvSMFjE9KRUyXkmS8PtDmS/pRhHbRljo76AUdlDWw3gBqu+TPEYabSisee56WeDfaIK/H6Jet/L1XtbmJ5lqk0xroqLALUrFpxX7IOw+uOawa5Wk1LAOU8qlZgjTRzWCSXl73nodGCuGw0PbUE44+qoQLFasTUipw5ZIp0z9IcDcrPlOGYuu0ZzTZEYNiC+Ms4RE3p9CBmAicojhTO5zCuKFWLOOldKVSF0q82hPjvUHaDc50gm5Ybrldsk3pKTwuyrqRhnGLYdw6ajHzy1RoxLiM24YOg3lmYrwVhsj1IU0ISN57nEGtygW0wFubWKFioEZ83qQVQ4mbNrdEpbB+BVN6LOWUqutKxdoP45K87YNEKw1KUg9NzfZ063Wf93KgSr6bUlahfprOPq6kAsI48Pjyq9SI3aR7pgKLly++7EklRU20rV1XgT7BAwXrvMSGSaJrYXe/aHCw0t2AmlfEOcH2jc4uQdm2EitBFQwqRGEjcVlK6oEnX26z1orKVUwRmjBbwpPbXVhnNOLTElYa3QdR01N2rJ+CDcvOhJh8D5cYQsfPvlQlwy1y8OTPOE31xwcfkxuB0NlRdI1fu+lKQyENat7vP/C+vBTx0N+mIW9cGsndTvFaP3fA7+/hamf3LpccKAdTTpNCfeBEoBI6sUH715oK2Bktpt6QPTMNbT9Y55WgCLcxoPVfJEzUd2bsLkIzmf8FZIi4Zw2+ioydENW6Zl4nLrKONv6LYO011g/YbjGbwRWp0oMdKMp+L1KFUKU0xs/EBNSQ2Xdk3WaIW+czo/W1S7FEtVtk/zGONIuepA2jV2h8BwOVDdjNsIobPstnv1fy0TSMIHwGSsE/peHfJiKyYIeXXLW492l0/m56fMdNFuYFmUqxQ6izXAqh1T9btqZ9QSYRQzA9DsKsdQ2UZZtWK1qe1nmhy27Uhjx9sv30IyWMx6RGc9pjiVVWCZ7KxUhtFR1lDIcR5BLMFZYq48PIxcXV9z8fKK8+msEeIu4jvIbWaMM1jDfhfoukekPbBMv6XbOTr3gLMTTma80dy3uqJqQ1DDd6vf7bMAfFA0DgIUSFG3tqBBlaFTUGGVqlvIphjekhWfXHKltIZ1wmanS5DtrqkjoOq2rrMXXO0vGKSQyowpW6QNZElUl9eCJMjKcapWdMjZGsY0iujn2VbU7w/xeo8L0/e+oU8rBnQrV1d9hncea3QVXZu+mZ9RJ2tn5Kwl14y1jlx1gPnEXjbWIKVQ6oSVmXh+y17ODJJoxZKrQYwlLZnD5Q2/+Iev2V4LP/7gChseOT/8X2yuf4z1H+L6gG07si5W1phoj/gDsRiaCaQU8USsT+S6UE2lSGOplWoMEsCKUzNyVeFh6AzD4Om2cP1yT9hBCwvVN0yo2NCQPlJSXBNwC2HnqUVTgpssepNK0wG9qFi+5LI+YG1NNEaPHYJ2nw68N+uPg7X6ebTvFXtjzPPmENDkmaduFJ4nr5o0ZJDWUeeB268j87Ex2I7OBR7GkZrX4xBKbLDZ8jAvDNuOkgOlZZYcsUEH35SAUOmC5erlNeIanbOE3HTYHxrbzYAPDTEVYyasW/C2YH3FmExnE960FadiyPLcTK8r+RXstkYltVbJNSPtaVW/3p6rArtklUYYY3DWrt27xlbJiunVLRnqz3RgbOHi0oNkalsYdoGuX/j2y7+j3n3F6TGwCa/54IPPOcXC5uqKfnPQgI0Gqao4WGrBibLA6/c+J2nfL6s/nOs9LkzoQyR6m8vTG9p5mjjgOzm+rOZSVXWvM6mmxwu10z79OsFaQwue1iLWQxOLiKceG+VuBBbaqM5zI45YMgU4jYs6tS1kjjiX2fSGZfoVXefY9NdMsypzxTZws9I1gzKQSg6UvGDbma332C7g7Mw8nQisq/9cwSjpcdg4tjuHC5VhY/BDpd9EYluQAEUqOME6Q25RTa9S2fQdXadiUOe0wwnOUUXlAzRFb9RSECrG6PLgKV9Ua1dT3pEYVcGIPP2ELg7WD8cY+/x36GeRnpnSKuzTz6MWoHaQHae7yNtvlU2+3TuqwJQLFNVvSdMB8pISTYTjnCi1UKUhTpBcaJJpxhMGSzWJu4ffEnbC9YsN/eCotVcdlmsYkxCybs4MOKNDZBUrrsSCaNXGYRPNPCFrdcsYfNDY8qe5mWjqcGuskP+nAbO+7GKMgKrOnTXPa309boLvHa1BXOZVQ6xzvtBr+OVmV3HhHefTHctjx+m+8dt7w/8ZO05x4LPP/zV/9W/+O8LmBqTH4ajGIqZQcgYcCUOh0YkaetdXij4D/18/s/9M1/tdmNarrR3T02o0rzYTeDIx6gnaGkvOOvTzXotSLTpHstaQU1zf/LoBaU3tCwmD226JZmC5f6AukG2mdBoHpdl1hY8+uqKFI6ZlthtHd7DcPh55vPt7MjdULhHpaFZ0jVwtJQvdMFC8oyaLrUJLJ1KaGQLsgsV3lW7wdCHQBcF34HyhGyrOFXyoKwa8IDVjvCJenAl4F4gprj4t6LsB0xq2WVgLda0r+qJp+KdZv4+NJ0Z1+04R3IAnxfBzJK8KFkHjixSiXyh57ZTKUxe6pnw0DYSg6RE6ZyHOlnlcCEZV3WmZmduEc4bDhx3KxM46yK7C0A8Yp591rQpZ085Xh89dcLz44IokE7av4Gasf9BC6/QYZo3aYoxpeKfdg7P2uQHX47IBbxFnoC3Pgl1ZW0trLMardmie5xXrvN5XZj2CUdbtsI4SlGZRMTgQFXhacTSxzFMhpUaKhrRE0qR/22ZnCT10a9Dn4CwlZz4cPOdt4XE683hyPP72f+Hf/Y9/x8XNT/jDP/83XH34BzQ8CSVjlppo4lUHKLK6I9bPUf7JBPw9vn4QhQlRl7yzTnOz0GGssbLOl2TdAq15Wu3pt61Dy6ZHEjGs8xHtwnxnyKlAH3ClJ9aOFgOd3SI+U23FtcYQLN5Xuk3guCQ8jsEFTJ85/N/svVuvJFmW5/Vba+9tZn4ucanIundPz9BIgxAgnvnavMDrtBAgHkYgMaOZZkbd1Z11yazMjIhzjrvZvqzFw9rukQ1ISAOIU1J7qRSVVREVftzNlq39v7ph45k//PGJj+cTD+/+Cdv9V6RyRypOrZ3eKr0d9LSTxkBUSVnJ0thW2O6E5eQsq7GdlHVNMztn4AxME8Ni6IqmyCyXCOBlhO2jjQHm7C9nkviUTzimPSwUHsyQyiABeVnC/xfKQiA2ACdqimxmWbk7o12Zu7jJEJvEgd7M0bHBwhgKpjFMhtCqc34ZPH2stDo4bU+kkvnql++Q1NAUto7zeaescXxMKbFthUFFNDCr0RyzdNt8SmqI/oG1ROyxM8jJWUtGS5iHcYLyJwbUGJ0+5DaYUce1Mdij/RhAPT5bjZpuJ5jUALbBRxzdlIQPkAyl5FnA6lOCEkmbrbfbBXz9HFszenVqTbTDJ96nlFUo20DUwOMBpmYUBssdrKvz5q7S3nRe9s7vf/st/+Lrf8vP/+I/5S//k/+ch5/9MyiPGIXhykAZDNJsZbnOpD+NKoLXPJjmMeAaJOPmk7nq8fSfed3iHR+JJKdpmYrqJhkJ1fBodWsMN3JayTrIOPXY2ecR0bnHl4y/+SV//N233LVn3r/X8EotDzyfn3C9cJ+Ft4uwLQP1AJO2u8y7LIwE+99/4vM3/xvH6RNvP/ySx3xCkrJzoftnjvED5pVyErYlImiXLCxrgJYuPVqFVdkvYStIJQcdrNNWY470CXSKBWskibVkRu/hb0tTIS5X1iaOFaoTiHZj9DYTFQnKGYEevjafRz0bdrvBVfWWoBiEToj53INFHMOREUyVudJN6EPY9875AimtLA9CbzXqq7hg3uneSMl4eO+RXy5OKbCuA9Sp9UBVyAvA/D04KTJRpsk22CeVkINcLZKigV7nkoPS1xRHxTmM9ZoEOWIDMgKfdIhceIdzu+BNkBF/r1zn+PVzGAnVAi2yluI4GkdF85AQWM+Mmqj7oNdOPabpOWe0xM9bipMIUW9kUoVmz5Sg/EenZGG5L2zb4OF+4fsfvuPTb/6Kv/pXf8XjL3/NX/4X/xXvfvaf8fjmnyIO6heGDBgXxAMCceIID4oxSK90BLzOdwX8A8DOf6Rnwm8RKDbNouYSgjoPYy2uoBmfZtqmBimhJcOo9BrNum4WRtsWtpX1/XtqzqSuNBFyLlwOxcaCSEJTYrtbcX+mtc5yEjQHm3O6h1//auP3X++M+g3jc0O2O0bv5FFZU+O07nAaaBK2LSNYOPLVcY+L9eXpQFPnmtq5SmYpCdOKYSylkJNSa6fVPkWYmZLzTDYcRKNMiuOMQ5IQNV6TFXymOsbGA22EnWRO/LDwdIE5sPqI3j4zIbHOoo05GH60sQ4iUsZFOapxOSJpdLu7RyWMynlxhj2Tc/SPrAWW1ViWyNAuOTGsk1KIS9MtwiNib0PhHZukzNwinUM7l+kIYMwNOkwxGzB/4wAAIABJREFUrbXbhbStK601Wqszt2vG4XgQW2ZXjdL0o3l8Du4af6/5DRRXIlbk8nQOkawLPgRNZeZ5Z3xk2i48fzrYzy2MyouwblHxjbQgJ+Y7VFWGxJCz6zWeEyZOXubRzA7KJpQivHujXC7O5/Pf8a//+/8ayv/Mn//5f8mHd7/kuDT+8N1Hvv73f02aP5/KF5zpNUPir3gwfXldbwKwmQXk0/ogNIvWCBeBIWTifN28BYCdHfGBJCHZZCmyICmRUSQl1mR4Vp5fjLcfHvBPL9gKcsrU58EvfvFLarvw9PkTjyk6xTQlUh5IGizivHlQslf0Z8rzDxekd5aeQY1uNfrikiMZ8qakHHois6D1xxhkLSElwKOQM0FKV1YrwFhNMGzgIjjKMEgeNhck/HWiINLC2LwURg+FuBDGVpGrmjhq8wSdTR9jMlLXTSjwJZ8ygDGEcZn+N3WcoMSXJSwqTaCPeC+1Rvvxui30fnD/PpT3UZGtnE4LRz0oC+QlMMN4yESLrSbwIQgJ8zGPR45bZVkS4cT7wpwx35MIMxpnDo+rAFIjkWI/dq6QSwzAKELVpAy3SESdCQ/jSqbYmBtIDG75Uc7RGH0C9oJbgrTQDWqFsa88fT6oe4tKFRdKgeyJ5D26cIrgaToUEgG++5hdh0zCJ7ESUo+UQs6gupNLZtuE08l4/7Dxcja++/S3/OZf/jv+9cfGvhc+PcPzNy8UN1RCo3ZrJpjJr6/x9ScwmGKuC7HWZw3xpJGDZdJMOM88ompHBR8kGWR3cqv0lydkDNTjq0g4nz9/ZFlLZDi3I3KzX75lkWd0c4TBMLh7EDQd7J+faNZY90gFePf+gaxnRIP9SWLonSENijm0gWhDsocIUZQxZOJilSQJ87jZSl5iuM7ac1Gn9WBpUo4Ny6fNw2dmqmqAuOHxC7ForS0EdRarv7tTXyJ1odbQSOER2GbD6D22EneJ1JBxHTrMDUYYPbCpNWdUN0YtpJzIWdE0yMVYVo/MIJxhjkqJDcN1yhZOuH4OfdM8gi3LINVxA7Nj+4oBUJY8afvY6GI4hQC0jx5DOiX6COe/JJnNxSGRuBqLg4T8shdcGcUQlV7/t6nwvlmDJoQwQf8rpiQM3KBookhGLSQXXjOjCa2C6sbny+A4BpdzhXbFuVZSISwsdlBy1EEFQC+UrQSTO72fKhrAfw8igSE3h0B4LDu9N8Q7SZTTIiRt3BXl7X3i/DA43icul41zvec3/9757b/7GNEpPwqVe80706sdTAFFjrkrxb9y0SlaG+SUp9BMI4ysN9wONDVkHPDyzB9//zXf/OY3fPzuO9Q6OuN2cec4LrPxImFWGTSEgRwHizPZGtg2Z/SPnO4TeiTOL+EkPz8529sYbEFHG2WBxzeZnIz9Zcb2XkV6LpAywohChBGskEqOXPBmlEWCASxAmqZXiRuutSkNGIZo6JJkpiUsSRhu6KLkub2krKHn6oOcE2nVSPQMxQApBZ2fcgzISJqM6NecBqXExmPGLG6IMD5Pp6k8DkEgstPtDN6gGxnFxgW5MnItcszR0DSphvla1Fm3L9qo679F9CZqFAUtsfFEAJqTNbCz7gMj+usC7xFKTpN1E0RslqHKzJYa5FLCOD01XBESOGNqPTRkTG/fjOuOvrYUOq+UElg0oKgJoxl9z7QqtCqcXy7UYwBK0jK/A8iZ2JTKxLZ0YhMOPozRnORKSnm+l2A9rcadUK8Z3mTcUhzJZ3+f5NjEuh2MUXFT1uTkVWDAMSJ+R66D9pbNFH/+tb5e7WD68Ss2dZvgbVxs8RQrCAn6oDgsyTn/8Fv+5t/8K8bnj1y+/479h08s8cwFMwLudU4C2YRlJEQt0gDE8CS0fbDc3eFecLsEkNk7ZVFaH9zfr6yrkkVRiZiNpDFENAsmMMRxzxyH46Z0j40gGHiNm4NQC7s7yeKGjO4xB1WEDK4cl4ERdDtpoMVJCzzcF3LRsCwoYd1JwcaUJerQaXV+irPlg8hITxPkdRk4AjNcTdRIKZTaooVEIudrw7GRy0FKJdIyR8WnmNF6j4d7DwOzj+vPSni9ZnNHHBGDHQ2t0/zPMo+Uc8OJTjmdQHsQGFfxLDDNs0I3Io5blNqNit1U6de/a/SrbGLM7HafMSVhnnYmmuZxw3t31NOX4eSRUtl6RCyrF3odtArt0vEuXC4dUHKe9ilxRHp8ByqYCq4aG6oIPjTcIgZtD2O15tgwhxmtGWOm+yTN0fhS4/hnHgSDu82crCWKLlrHqlO64AeMroG/eqjNDYnPbX4vyiB6gF7f609iMH1JF/gSPSFa4uKzcKqvknj+/d/xm//lf+D5+z/C8YIfBxtj6ulkiv8CH4LwV43aWXNCNKp5hsSgeHpuyJbIOSwHpSxIGrx9v/LwRtD0Eqs/A2QE+1ViPVpnPMvLUwdPWAfxMOSaOXUeKZEQ7YlF28VxeNDtYlEbbYXjMHpX8lI47wdlVd4+Lng6yItyd7eimikl8CnNc8ASgLpkmcpjj6GWFOs92oVFQ7iaNMLP8BliB7hQbZBT4uiDZdkQhNYugE2PnKMpR3wxaW6508SK4jYCkDdHsk4hZmwo4REMjIapuYlyiblMWFQP2WQE0XwDwm1uCy5KPaKxRCVqvZNcA+ICvFaZh/d5tBsjttcxrsJDporXken7szYnhgmtxs8ZTGWeDJ5zucTWzuiUDI9v1ikVGJi3mf0VmFnHo21X9HZB5yleVQ9MS+ZDaQw4asTzuAVb6F5jY+pCr3swf0noI46gmiyy7U0ZF4Mm+BH4m7hh7nSINEz/8ZZkvNbXqx1M4gUkQrZEEzLpFy3RgKFRQUKSlVQbfP4jv/uf/or0x695sFDn9i4Rr3V1xsN0gse1aB5P+CqKzOiQbh1PwroJ6wbVFKfjNEQCw0iqUxcjU9qpAbAzL7SUYAsWR2XMuNmMjTTZt0FiwX2El87BKXQi+kRSYVyEY99BMw8Pb7icjVLekvTAhnM6bSwpkyRT8sayLnRvaBEGhnmnjTjiZE1kCUVzSSmsnhZNK4gwbNCH3Z6kg4lrWRSCmkeGkCBgB8d4AY+EhOwJTRIPCg2JgyJ4l1nfFEB7mptcbEdx/Bg9AHxBUSmIGEabgfxbSETSl9x26xqtxYAS+JzJF0WzarpplMzAqkUbrQfmhirSYstFRqRuqmLduZwbYzbv+IiYXx+Jdii1Gq0bWRXBcDHKKSE5kjtxY1niSnAL4sIcvDtiHrrNqoyeWRYYXgOPG+GpziXTPdFrJFD2Pn/CMYdVmxnq05vYjbBj5UL3RuuDVDMqQZD0lmjVsNJI+T3KgXKenYo/vste57YEr3gwRYDx/JVp1HUQT9d1Axs9Os5a5a//13/J+eMfWVql9euRbQovU2wsIoltXUJq4Ib3MDn2Hk8sJKqfSBIJjGJsWwHJaBqc7pR1i7OD+LU+SkNTNKNNo7mlkwvc3y+si1IyfP7Y4u9qAIm9DpLEb3eYVVIyfU6dcGkId/eZ3irDjPuHlbx2Uq4sa0KzB12eIxMIQvsyxMhFGSaQEjKuaaDK6BE+l2ZEcbzmmkIAr9hMap0xA1ftTgDvsdWEry2wj9Gvxt4pXpxpCQCj9wjsv4b3ze1k5kJQ62ApK60XWgv3fUodIboAxT3ytVqLn9EM8fAw2oz/wCMDiZlk6mYhPiVFW65dGbYQSyJEBfgE9/dLZ3RhMHHLaR+x3ml19tPhoIFz5pwhO1oS66qT0CA0ZO4USSRN7Icw6mCIcN4NulLdQWIYW3Q+xRY23487t65EjyDQUNhb1IMHg5huBa/tMjdBdyQF62YeR3PmsXG++/8L1ffrFVu+2sEUQrDbP4AJSkZl3sA9hlJyuHz+no/f/pY7iUFTykKt4exOEvGwQuA5uNNbFAdeTb8+4ubKS1zOo0dkiPWK5mj73Ur6YrvooejeTtxEiikJzR2no5pJ7mzriVaFnDKjK999u8dK35TCGhemC703ahsMT7gEzuBEeL66YK1y97hS62d06SyLzBuYmRYZK7l5j03RW6z1cXcF3a2x1o9haMowY2REJTYJnwJW6zDlCEIYdZMGVtNHC+uFAhhZwX3Qe48olDkAS07stYXRV+LIWGtoicoUO+YcNH3JJXRidhdHTA7SMhDtN7HjaP0mmhy9oZJveizVFDewRwZ8Sok+QwAF6G0QTheZkocYkNUdKxFV0pvShzBcOI5I6Zy8BaVklmJohiROKRZHJ4W8Jtb7FJiadWTGv5Q8ZQXFGIfQSXz8puENVgqi0UVnkyQbA3Rum+Ei+XJdQgTwYdG5oCR6NXo9wogtQtbIrDdkDugpFJUQolosrf8nEu7Kz73G1yseTF+sDnhUe2dJmE2syGJNttF5+vg9WaY1w/yGWaxlo48WS5c7w4KhyikzxqDV+Gch2CvJKfJ8RPExn77TT3ccO2VZOfYW5tPRGc1RS+RJvZmGLUOTTfPxwXY6MXpnWYWywHHEJjSakVxImmltEAWthWFCrZ1SMmnNeA83/HF5CsuCjMlkD5BEKgspx9PaLej31saXnvsrQMf0GfY5nJh5Vha4k4pQR5QPXA297obLiFTFVtGZzOjdI/ebHxVEmEcuU47Ndl0XfBiu156/aynp3KSi452UMj98PPjhuwtffXgbDSBUzGtYW0bQVynHkbl3w2SQy0qWRLN40JQc32kkADjSBZozDqGeByWvgTP28LodR5s4U6LW+Kyah10plyBEclHyomh2yuKsSyYnGD7o5pRNwr+YJrzg8XCLTPQoKujZSCRSgcslRJMpgWS5kR0QgyTn2EJHC/Go+FUndbVcyfRIRua4At2cbscNj+rAMqf4eirsQyHLzND/Yjl+xYQc8KoHk96muSBTdTs46guScnxhJI5LJeeN3gM0vH3o/iUYTpJMinXSzk6A5hLAbLBUYdi0HukAOeeolV594iIhXtxKorUQwLUD2mGkHPEjJQuWp13fwpRa+0tgY1m5f0zse6Md4EPxoTTvccyQQquBO/QaLNFxbpzuginLq7AUDa/bpPqZ4O61dRYJ+0UmkXKijY56YtRBax3NhSv2aWOSAio3bVEmwFG3q6MqvgPrHevh0BeNJtkYRrFZhAgx+vJ6G4wUgLENv0UEu0+Da7dbLLLOuqfHx1MUZWrgN+6BKfUe8gCzgbWJDdqk+VuNCBET1ArW4koJNqvAEHxXxg6jZcYR35tZZFDtF6d1o5Q4hpc1gw7WZYnjqkAqTiqD5SSUJbEuEHNX2dugrMJ6KqQsQOHYL6GxUoAxB5qgXTndFy5PLcSfKeFtBAjfnd4jIfMqbmVMjRvQm8/jqdyGeccYPQZNmuyjeth5bFFGObEsgpXEafs5y923uHw/k0dv6Mirnk2vdjBd7RHXJEvBaceFlJ2jnnFzWheOc4fmuGwgC6o9xITDI0wufXGDoQkfXywIU/Y8wVCl93o11qNklrJAOaZaOk37QoobV5Tj0sgaIfQpB84Q7SaBHUQBpEUDyhoG14fHzOWl0bujIwL0I9POaDUyyHFlVKMxKNk5bUtoTuxaqsjtWNLaAGpUnuPQ42cWBZlAajCXMhtcE20C7qJQchiZc1JkHktRwUk3vEX0KgyN96bz+CgS/rk0Wa+jRn9b87B7jGFzi5HblgqxJcXnGcLQYWfWbQGx2JSkzwcLUwlNfCkpChgCOB/0SyNHdg0+yxxb78hY2V8akT+cefp0Zj/qTMaUeZwUljVTSgygZfHYLGioCqkoknoMny22ppScVAILWnPGcUpZCKzNSUmngTqOxa053hQfiW3dqMdH8IxIwaVRD6ceRu+wLMqxx7ApsgYhUTtmOgHtgepV5zSP8DK/C1VO2z2+GeVh4+7NL/nJV+9489OvOL3553z4F/8t5l8D/zD25B8H03/AS3wCpVyxoIgVNd/ps3Wi9cRlb+QG592pz5U775Sy0L3DHE5lhpppStRao4VXAsB1QvY/LLatNKMwLpcDF+Hh3cKbN488vXzGvbHvB8uqLJrIulCPinXBuofKW8OcKQ5IeN5UnVxg2xTeFawX/nC5zGNF+NhqH9QKOeU4gvYWcf1zECkRjHftaYMvjGAuIaoMj1zgDL3FAN7WBU+C9TbPDR7byqTno3Y9RI8Qzv0x6bnIio7hbVPvkzXPZMpZftDnMVEWcgqUvNaGm9zAWwPcRhxDcqZ3p02cDyIlwmwPKwY+jzDTNmFTo9R6RI14onen74FrXfZK0jwHNVz2weXpTD0HtpRIwTiirFshTaFnTmEMzotRlqDcU5rDT2E75WmTi3SGlBR03I6tp7sTTontbyYRJE2kBFdAyBr0Khwvg8/PwucXw8eA5/g7x8x4d1P2I6woqsKpAJ5xU1pr83sOCCKwqbjGEJ0G7o385gPcK+XdG+7e/4ry/hF7fGRf3tIs3RjXLzfY/4c37/8Lr1c7mDJHgLEkTMLHdOw7mhut7+wHHLvAYeyfn3FPfPPx4Ku7zL0oa1lAO2P0KMI0p45O7xbWFEkMD2V50hReJglXuQAIHEenfmd8kJVtveNyfAxQcjrGk0dEyHEZ6CKUTREJOUCeFhSIbCBNHS2Nsi08vit8/kE4f1RGW2k1AGcljiHf/jDbSd4mTnfhjVNv4QXUHFuNGWadNi6459gUkpCXHHaGEUwZKcBpMwscJsCswOlSaJ4iLE7B4+nrNZTgKQxh0dBhwXId80apdaCSMcvkVMhl4WiXGEDdqN7JaZIOE0NRWWjNg8TQFEdYosXlZqYdKfAVic+yjxFJBTWT9QQtoSPh+6Cacz7viCSOI2QGrQ6sFTQV8ixb2NbA5HJycvZpc2FGqIR+K6lQSsGJ8gdVQ1Ick2zEtbOtgjUJVs0TKWculz3YOmWmX8TFI74iIwZ0fnjg4c8eeGgbvd4hbCSNh8IdURtmc5CrKBkhpxK56/E13ZIhxsQL06wEV00hyzgttFQY5ZGqD1Tuye2euyQUVoSBS8E8gQyEHnlRr/T1at/ZDGflinbItbJmGB7+V+ql0i4NqR3yyvLwns+XJ57OO29PC4+nDVGjTqyj9/gyckq01m8izVpbYC0Eze098BvzwdEPEOcXv3hD0oxII2en9YN12xh0Lvsg30uot0UC2LymF3rgTRFFC9hgbPDhZwvHubGfd4ZlrCdqC7vD5WVQstBbDEEFsi64dXo1xEtsfSp0cTxFZKt56F366FOopYRJJJjFaoCXGJwpdEilBE7TPY4R7TDMQ9/ENMXiAciHyKkE4zOUFrTRPEIFfuI2YujPaiwNlR9Cx8ZgVKd7j6e/cwtec4+UgjGiBltEEQv2rffYhKwLYzfqeQcr1KOz7y2yp0RnqF8COShFySWaY1xmcWjoHILwmJgcKco94/jaw3oyQ9qw8CuKZfDEcTZqrMLYUwtg3AAfWNIZuz1jX5oxujJ8Ia/3pPKWP/9nf4aNGEzGMgXCcrP+mHlsf52bXMBtYmBZEfFwF0wRcBAaEe+DNrobut5BVqhC653vv/+edtTJ9r1mHu4fvl7tYLqym1NeE3nKErXQbW8ch9OOzrE3ZDjrw1tO6z3j5czl6SOfzx/ZXzprDqUzZmQJffIx+vz/jyfQ0QaSCBwlp2AEBRBjFOf7by+UpHz180f6eGKMhmBc6oXTfUG0R5W3ZzT8EQwIXcotcC2OY5oGZXXu32z85GcnfvjumXpIeK4awTqa3hioYx8RJOc55BESOJmk0MjgIGtIIsw6iYSmSKe03jFlVgkRx4LZZgKThncNJnOyPkKaCQfXJAcCjPLAw7xbCDmc0Em5MOSqg4qGGhOhHXHkDOOsI9YRMvUlfu1Mxs0XEhs2jNpGBLJ5sGa9Qj2My2XMoQf9CI1PKXUmi0af3LLkiYEZeSkTPG9ojiovuyK+qshMuASjjcps9qbVa5ln44okrGukG6Q07Ss2EBXqMPJi6GjkWMEZJnRLIIlhhXX9OZLvGfoG8luW7Q3mW9SLLdySDVrrLHNQB0CdpiRC5mCKdl+zFiLKa0UWYWh3INlC9hwJS0siq5DZULZJQsTd9Kcxll7xYPrxaBK5xsCGl0kloRg5JS7jYLQRTx7PlO0dW1pZHt6QvdEuZ/b+EsC5G0WVJS0M6xGbMjqDFMWM3VAj0gpCkQgW/W7f/OHC6W7j/vEO4ylc/KocRwuQ+4BRhawFLEymV1GjZqHXSFVMSSbw6jy+y5weCp9+GIya6SNW9DoONs2MUbnsg7ta2M8Hp4eQNXiPbU5LxKHQ+81KcjWuppzYUibEoGGdSTqmZSKeyFkM5gY6xvSLacK8g0ROUwgSv4hVxaInTsjYUFoVkq5cVcTuPuNTSmygovQ6oAm9hws/hhwcR+PycmD9Qq2DYx9zg+IWyZImyBu4eWy1ZV3pnEk5jqE5RbZ1H7PWy5fwIuaET+W3jVC39zGQOmbWE9PQnKn7QEYA8suyIprovXNB6NbRHAxhryHmDM9ZQwgSIi8LWjbWu3f89Gd/zrI+kJaf08nR+5dWhkVCg4riPEebMJBHtM64h0jGPCAAG7PLLkVWVUIZoyI5h/Nwugyi067go7CkyRKak2QD3dBrqgJ/KvvSqx5M19c8yk1VrKhOMLbhI9ikAbRhSAsLxkAwSXQRdMuc1ke242DUA+uN81FpfdBHY2YAIFe2TuLYl0uKjcECVK5t8LvfPfNPT+/Z7oWjfaY3o6TQV/XzoJ9gzUvEeGi07qp8CSoL138INct0nv/Ff/zIp08fuTwZ3kMN7DkC7rwO0g77LmiOp+Z6EmSUyXY5jGCKwv8X3fY2jDam5+0K6qrgkmc6gZAAsdkD54JVaC3ofG4K7RQDwWd1kxu33jJWWk30qvSWcAsfGy68PJ9nskDYcUYzzp8Ho0VOkc3tB67q8thmIrVAuUb2SgJZoi+QMW5i0N46TIuPeMKmQNExhgmJA1Xjbl0573vYkJKwbgvDoustpRC2PjycuL+/54/ffOT5ZZAkcX6ut58nSigzSEEMxijxHcXHw93jI1/97Of84ld/xun+HZruSfmBPpSGzUbcED0ui+J+RKyJlRn5nG4GavfpPXSARPdpeiauo5QE15nhPQkJIfRjklLAgBKbP+rzXlluRR6vmob7P7xe7WDyH32KVy9U0NaKq5GWRBlC3VvoW5hqWUJc6Bo4h7uSDDStpLuFNAbbG2WMRm0Hl+OMDafVyugd6zO36NraIXF0EFXs+8Z698xf/OVb0APpFZiVzArH504WYTElbURrsAUzCHGkQkCv0S1+8NNfveE/er7jt3/XQd5wHPD7v/8jnfBDaRaez5EFtK0njpcoUpQsrHcFLxk7rkFoV4ascG0u4RrXoRpwF/HE3/JCRqaAMeOH0/fI6BaUVju9K+0Im0NgTU5tdf5DDKPvv3viOKbRdTCV7AECW78RgbMCOwMR2xHvt08VQDBhIGFhCZcPmoTW+g0XGhZyhD5TJJMIdfRQZOdoyR2EpEITlJPzsN3x9HQmF2XbEik7IpEcaq6UYozxzOkO3KMCPmXhuHQe7jc+fzrYRwvl+ghD8fv373j74S3LaWF7uCOVzFKcXp9ZtwSWKLLgcsF8cCoR8NfbTsTeKKInQMkzuyrlMiNfOuZx/HWJdACZoXwqCZlHY3cPVHwmUhpEvoo76v2m2KdcjcMzUE/+NHamVzyYrqtnfJjX7q5gTZhPwELfO16NfKeYtvC/pYSJ0IeznkpgM2bkEACFQprMiY3FHqJNZBj7Zefl5Zm6H9QWLE/kyE6BJfD1b5853QuPbzJKMFRj+q8uPrAxeKsLwrhhPDJB5DSPHGis7EWMPp75xT95BC18/CTktvCVfeDTd59mceFVnyNczoKLolWRlGm7zohanQbWdMMlbCi1DTQtccRZFo5jD+woKyU5Mo9qYzTqYTx9OtgvHv6y4dRLYDxpYlZho9Pbta06h/L8nnwEU2TDgp0yCWC9hQF6eNTuqiquA51AslvINWzEUTMSOwUxJRUjFYlGYQRPTloLkiuanW1TyqLkEjlV26mwbDk+lwS5wN3jAiJs60IfnWVdyCWU1kkTx1GpR+L9T9LURSXaMah75+37lTYCTzv2xmiG6GeO8ZF+CC/NyLnw/LSS08pS7snpxOjCx+OCJ2ddcxRoekOs01rFKHFUTSUwJ67fX2Ipb1DNLOs9aZp1hxmnu3uQ8A1GUucILVXOlPJAStu8FjLrtqBacReW2Y3nM/0Bv2rAX+/h7vUOpsnmxDEiUgnzEu0PJRkpFcbFsbWQBlQC2PUxaGMyLMvCDIAlm6MjjiElJVptYWhtDbFEb4Nl4gHoQuo9vGdSo9UkCesaIfmfPlfu7t5ydwetnhGLyIk0wfVRDc8WTEqOiI58reD2RLcQbGoWyubceeenv1j4dPmEyCOnt1uYMY8n2qhkdXoWPr+c2Y+4EUPPY/GEHXGRqTj73qi14xabi8sFGxF4H0H/MjOXvlyYEQYXwtFw9ERDbJ5P2cu5BuOksZU4U408NT1moY0SDw3V6J3R5g0gaeZgzc1zbjpRn+0Mj7x18xhIKOQ1s27K6Q18+Okj6yZoChHpclrQrFQzTneF06lQlhBJjnGgyclr51r/fu0eXNYFOFN7Y1k8KtK9YxinR2W7j6gWlcFobcoEYHTI+cQYwuUcrO0Yg1IKEY0ssZlYpe0vZPnEIgtDjHQHkpSskW9ecvxe30LD1HtUYDmhBxMJc24/QpTaj+iIQ5TjqPzgEsF2XI/Cgf0lTdQD+uiIb+BvSJuiyz1L/ud8/OZ3gXsKoYNzQSwh2oDl/58b/P/m9WoHU0DQwZxFgNlsEiHW9hASKtIL3jpZ8vzAjeEZ18GQEb0EKUEb5MnuYRFh246DpsJR48ZM6qTWMVXyDKaTclDWghbl7j6zLJ0lV3btJBv0PiiqlFRmy8ZCawdpGjRtGKQZqu/RWKKuUeAo8UTHjdO98YtfPvDb3+5YUj7YvNFoAAAgAElEQVT85IF+gYUOveHWOT8dXGgQeDfXBhk832Jlj2NMWcNMisyVpcy0yhSs1VXDE5DNZAwlgG8hxIxihFZKCvtHp43gdCTPsgMJ7CRwIUG6477PITed8BabVB+CUm5C1uSDZcmUU2ItSi4hXlxW5+4xcXrI+GyTyUtlXZztpEgC0x3NwikpyyqkPEjZKSWO/mUNo/dVOHvLXtKGGyxT1DmGz8bjkBr0PshZWHIcNUeLYDhBUSq1HZQloSnTe5yaRGBdloDjzBj9ejt1RhvkWXFVcpqRKc5o0aSya9hhIuuJmXLaKSX0W+4Bgvd5fHy4j0120r1TYuDkFMd2t3vq4VirjPZMs539Oaw7x8v3IY1gTNU5BLH6esucXvFgmvgrs62COPuLEPGtLY45YgkfwYRZc9SF7pVBR8oSaZJmcR7vkZ/jfaAG+abpSbERdKPYRj/HRqbqpAJlLSynwnIXzbjrCayfufSD0909th9h4RjGcURuzmo6dTwaec5TghAAZ4DckmwmHnYw5T0LT0+dz8cZu3TEKkfdgzKbXWVRMS0x9TwHdex1SgCM03264Usi0ag7RmwIkdPtMwY48DOb+qVbMqR1nM4wWDUhvvDOM19//UTtkcQZlhedw+mqQ58DyWQ6/xNwravOMYy9ornz4c3Gsg3W02DbjJyNvDjLJixbJ+U9jNAJTncbJQNap/BxUNaCZ3CtkIQ29V25gHeNbrmpews260vTSy5XY7CwLFGMYDPne29tRu0CLhhO743eGkmUnKOhZr90ejNOdwn3ME8joCWyyA3Di7FAbMoSZuokiTqz2ZXAHPX6dRJezqSJ5GUOTaekAK3dA1/VadFxs4ADEnQzXC5sm0FOjOSY77RiWLvwcIr7Rnz8Q+z21Y6lVz6Y4hVokyahlEgylN5gSbQJkC6n+CJdxoxHieoiSdHtnmZKo6uRRLDWw9mfw3k99hb1Rr2z3S9ojviPUpSSlogIkQhDyzmGyrqu1Po8201Wvv/jC2/frOy7sS3BdmWbNgIJbAU8aqZEZ+xIDD6TAd45mfGrXy48ffuZVkG6h3G3KCLhnI9q68J6CtPvcbTAYRbm0zYYqFBORy3T1aMmEq0muSipBAisxGah6tioiMO6QG1Or53eX9jePJI/FuolcX4xrHGLxlWNP59S4u3Dz3h6eZkiz0j2RKLwYF3h/kF4+5OVn/9Z4e4R7h+F06ZknekBKW7WenQue2P0wVEbZiG4TFmQnGkXR5Y+M6e+pEq1HvKJnn2We86actFZxwTeohnFzGm1k1IOZlKESIG69fDiFvKNUY3eYvu1ElKLdV1JqqhHDK6ozrRQxyWB5Khtd0Ens2bNcE+YCvQZVMhMIpikWR9xpCwpJAE2vmSXpyQMb7GlakgNjuMgpzLjTkZE+niA6CUrx3EwetRGCdwY0NdO0L3iwfQFAxHkdvFLKjPHKHJwdMkzN6kgEtiG4LcNgnF1tQ9YQiSXZWYCLSvnvqPZUGusi2HDkdXJKaMqFA3rSlpSCDRd2NLGXSl8+PAXvPzwe7Id3JfOqIOLDCiJplPcuCTyKVz6mmSmPSqmyiASOjUryyb0vvPmzcaf/3rj9/2MjrnJzNUbkZknNJnHeJTSRuSCv1kWTneJu/sF0UHvldYbWWIzHKNH80px0jI/U01c67GST+/gDMdv1bChOIX3/YH2TSXrKZpQJr0dMgKnmVOTMFbFcofREDV++ct7fvXrR376i8S7n2ysp8p613B5JuVBKUYSIlXAHNWV4ygse+Hp44Xj0qg1lOqalO5GXjN3STEsbsipv+oeQL0JUwjpnE4ryIjIYghveNYpbIwwQCHkFn3GrJiFbCCU2QO6oBbJkd4tjuuTiew+ZjZ3bOUiGlqpGdUMzNRQpR4VHx5sq8UQHr3dxJ8xOAZVduizscW4YYRt+g6Z5QQpBRkwNJp9YylfqD02V02Zlz0KIcJypde/Zd5Zr3c8vd7B5FMXwxRXTsxGRSJe1RKaoxCgWASTybKSkmGmEKG6ZIvjUxbFrM9jXQqQFocls0qnCFSMZgHKioQNINFZS8Kl0ltjDGHfhZdakeNglZ13p8y7+weOfqHJQe9G9sKxd0pVKIH/xJN5TF1LdK4ZwfRE7pGTbfDrX72hfx6cn4/AoAhnU2xcUbukoqwlMbQjA+5PG3cPhdNJSaXTx45kJ3tgTUUThSgk0Jk1dM3b5pqkaDFoUg4BpCbDR+dy/szjV+85yKxW8JldrfMoF2rpaAb+KY/0/sy2wk+/ynz4KvPuJ433P7+wnXZyDnyrD6W30IipdERsRvy2SA1djLRkLp+d83Oj1zA1NzPSWmi1zfyiWS4wmafWIyWg96v40jCinbePKF44qKzrQk6Z46gTKI/4EbsWXSaNh1guuI0Y8CksNL13llVRLWAeueAeKneRyHhvtVHWwLeGRuX4aCOA/9a5zE2+97AFXRM/hQjcS1NwiV/TRCHlqHC/9v51wtupEKRK6/SjQc/xoOiV/cXoLQSb7voPUCX/x8H0H/b6cSdYeIIKpUg8ZUzQrGAdXUIV6yMuKDXjGsOaxEgOYoNqR4C+6gjRxir9TNs/Ya1FzETrIVpzZ98H6qFnar3RjhYPRokn4NgbWQbHXeLDh3vWu4L5jjfneBnIMNZTY9ky5FBUp5xDNZ6icDHqu4VeIwVhuUuoNX7y60T7XeiMQuk+FcaakR7qbXfn3bqRi09N5AXNijMoRThmjnkoIkMwKhqYm/84DncmL4x6je8IoSnE8NhO8f59STx3JZUMNisnR2MpATiXcgAHpTTevV14fIR1O/P2vXL3trIsZdowHK8Dk2hrgQhoEwPpMQzTfaIyIsl3CPvEVbKCtUbzjq5xZJ8M+A166yNU4KUIxyXM4Few+qYkP8Jyk6a/bZiFPGEOht7CtmRjJlNKmQFuYd+xDud6sJaEtdDLjQHiPn1ykEZcK60PZJ3FAsIcwka3gbvCiOwpJ4a9krCr9Wi6EMwH65LBIhfsKiHoI/KubFRkQDsrYgdtdJwN7YoMZ0pX+bJ+86NfX9/r1Q6mH7fvRs+YgOvN9IgQzFbpJDs4zURAtU61M85Or8/0emGMTru8MHrDh5KkcFx2at3pveK2kxRWEfIpho77PFp4D7xIjFNJ9Br54L23m1L35Wj4pyd+fnfPtm20ftD3OKKMw+l7gOu5JCoNdcgpsnTCgyCUkmAYdRykk3H/lfPQlJenMMwyIjxNp7k2y2yJdQPajNuIg29OhZwLSqWNFkcWDwtPMHKhNcqaSSnTWseHUHSJjTIooTgWG3Tb6e07BOG0FlJJ3J9W2vGCirGUULKf7o1cKsvm3N8b2ymOX1oO8pKjI2+G+WvuFJ2ZTnPQ3o6WI7K6lyz01Hl8XBE3zi89jlgGako9hxgTkcjwnp9ArTF0DJmG7GC4Ug4ZhEiA4cNDW5ZzJgs0D6Ht1dcY0oA4BpUk0/rSo6dPp2m6OT6U0QCVWQ0en6F7os/yhTZ/XdYQSBaTCK4bznEJFk6SxlZkLewzRwTxxfE7lPeRXw7tmLVcGkmX6lAvju1B8kQ6cMYbiEWihsiPy8h5xZzcKx5M/KhaRiS8TuEeD/e8zfN1Mmd15+Mf/sDTH/6I1CfQSuufEdlZspEw7pYIjTdRrCvqjeSNVOK/v+I1JWfKstCOxrUeChK9Jo7dOdzC/HoNj5/RILUN9r2zbGtsZaNhx6A9FV46LA/C8r4gEn4MH8aQ68pu1KOhHmUArp11Fb766Yl2nDkue2Qu2cx/dpkWDI+NcWqkbtG2RCjcmHG2OetUsROZRyKhIiZUykpYUYaH/CESJsMtr+qseaBjkEX48CGBdFR3bKtoig1rKYl875TVKBssa+R2l40pzLwWIhA3sE06IPqU4r1LDGxHI+fJnXVTRhMeHldUMvve48/2WU+kExci/p1mH58b1H02oSSfeuvE50vldFrCGzhbZsa4lgAMhk3vmYS8o/Ue/sBZSAqEuhqjHj2qvdJGH5XLJQSNODPLfIpdr6JeDbwoxJGR1xSYFkDE5XYiZ6qkwuIyxap9NrY4mIUXUWIHGn61zRiMhO9CP/z2EKgtYmh+zMX9Kbxe8WD68ctvMQ99TJr0tGJtR2vnu7/7G779V3+DPJ15cwLNjbutsp2Msg40WfSmpYjXsA61GMe1e6tMVbEFlS5yIBp1RUUzvTnn0W839bIuuIeyurU+jyPO09NOLollzdjYYYAcmZzXuMiOM8tDDhaOsLSHTutq+bhuUMa6LBTdqD8Rvv3mJW5ASyiF1i+EIP4a1haRurV2Wm8RlNd7GGJFQoxqkZt+xZGwmVOtNnvngvXBwmOlWUEinRNx3m0FEWdbn2+13nkJJs18oLmjRShrAOuaG6ksuDZIURvkOK0ecZyb4L15HDOu9rxr91xKMjPaZ/37cLa7wvncwqDrTs5KrzDMaToD+MPdGpjX1GilKRjFhWUpoWGamE0fPrffyZKNwB5lngvdnUUzVjvH2VgWBYlCgMhFTxzVcCtha2oh6KzJb/ol64b1+AxyC8yvDosSgon9RH9dANoUDeMwckth9WnorYfBFMMKEZo3OvQB1pXsK3Z4+EfN6UNu+VPXMo0/hdfrHUxXqper3kJnAFt4vfw4EDM4Vy6//4H+zTOf/v4T8rMTP/mwcloyCw3GHkegaf5FHMmg5mwlhs6QHituEnymBgTNHZqn4YOUhVyYCuegtEsRci4zHtfpFT5/Onh8s7CscRwZ1bA8GBcYdNIqc10ft/rvCGgDH4l6tJkXZCxb5v5R+PQp2lWENAsXr6beqOEupYRS2WzqnCKy9hrXO4aT5++Pz9axEUNnWTPZEmZh+r1fNVz54uRcKIvi3sipkxTKGlsU6tEqk8FlqvNVIqZ2K3SPm2kMQcwDk5LonBsTv1EJXY54DEQznxE1M9bXfQ4wJZfwlC2r0lqPRpmh05YRP5sj1B4PIncnpynwNCOncKmRJqhsTj3qDSQPAB9CQR9mPZtiVB8wWijBsdjgbhlKw4iIGue4+MzihnXJDNfY+0eGPug450ukTgwRRpsFAT7blP2q5Z6fF1Ako6azPAHc8pdk0GG3mJjhQj2MNQmjJxpGc2gz5+kKjPzjxvT/9OU62aIZ+yoW3eyjMmSgQykVvv/b7/n2337D8qTwkvju64P+Mvjw843HrxbSXQDH3hxmeaJ51FeLTsajxwWuGhSraArKF3BVTKMjflkVRqJeGjZijU+50BF6h3EM9pmto5pJAv3oPLfBMeAhpxCE2gxy81i55dqhp5DWhBMga/cX1oeNNx8Wvv9mZ98r9Yg/d8pLBMl5vOcQoGZab5j3G8uTipMXDTtEjuGlGlvf3f2Ce0SBqOZb2qcQ24NkR3JHJW7yJH6TO2gWon04WmFD7prQFIrqPiCnZUYJxCBm1jEFMZrmMAucq7fG0aLtgxFbABNPsT6wcVDrLEQQ2LtiHVQzJYVh+xrQ30cch4bHkS8ppBLaomB3jdFmxZNGuJ95PPDiFDc3S0JEOpojlhET2gU8x+cGGU3Cvld6c+oxiQoRjt2YKxUMI2uCHkexK6cjzFA4myWqquSsDA/dUbMxA+KE/QgdXZ0EBRbb4OjhZug2cMucz5GoeXinS6KOOhXsf1qv1zuYbi2hU7UqUe54TQ+QOki78vt/8zVPvzvz2DYWuWfUzvP3B70+8fQkLG+Vhw8n8magnZwNXWDdSlyA2bl08N3QpZDzSts7OhI5LXRxqhu9hvG35ETJMFJsST46mCAO1iMN8fPHg8HCuzcb4pVjr3RR1jcr7TCWJWwqKjIDLiNzyjSKMvsIVbiNM9jg/l3i8rTx/OnC6IpIoh2hBMZniqVbVPpoONnXJfP45kRKIwSFbvF7pvhx3RJaIlGhLEzqXRAd6GT8ujhS5gLJVHwDEBah6JsLYNVxPGUGTq+N0STC/b1HjdEUmIrOwLUUOdo+33urNvU2xCnXBO/XCOSrGjt+hru7BeyO3/3hIwkoOigaA00FJCeqhVdxKQXcqEMZR6ikryrwmzocQiM28756mlvcfF0tT63OynCPSJecdXr0evQFdtBbsmQ89HxuQYPAFb+084TZJQaT0swYrSLEUTsrFFX22V7chiN5zIaUqbPyKDjFI6fdZnFrb4ORha6BG/rc/K7305/C6xUPpnkhE19g9sTiC+LG8fmF//G/+e84PSm//+u/JfXOfnzCOqGAFjjOzlM38pOxfPfM3UNmu1PWk5JyZdxHvU7OSr0UPn8UXDqn+zxp9YyP/529N3uyJDvO/H7uZ4m492ZmLb1iaZCgyKE0MpNpedB/LtODJBNFs6E0Emc0MwQJgntzgG50V3VV5XJvxNlcD34yGxy9yCSaLMeMASSANutOZOWN8PDj/n2/rzuatjuhoNRGxxi1E8RjqGtzcFqtRu/e7tvmefOrRtKipJwovXD37kxYF4iRJB2RhmJgMsMa/UjUJ+xZIqg2klZuPl349TdGq57GUimkrKS40uqZEIWcH49SgcMpczjphPwXVxGPQcyuVjApECBnTzgJQV1PM1uV1nw1r+rH4DDJl6O3mcI7S1R3ceUYYFKntSUiQNlcDmFjRlObEwrMlFGNoS6GbeoPW2+PkiovRN1cyCnEiRwBzTAsYiUi+YrSI7UbuXdsVFBopRE1sLJwboIOmZ9Xp9QdDXH+uRtjNJacqKUjphOh86gN8i+rEeu+RXPRo2ueRMyjwcVnPfA4I/TOKCW4OkUOR6W3QauBNuZcfQay7q1DV6zO7Su+UFFzT2gffX5bIwclLUrocDhlB8JJJ+XgrKpgRIlQwMZGkci7uyPHn1X3gk4GzRBhaHdpxDO9nnVhejwZG55CWmoj2WD/5o6/+Vd/SX4P7IVldZhYrYb2RO1KbRsE1+BoBAlGiJ3TVWQ5rhxOTqZMOXLp8P6u+jHuzeDli2v2yx1BBtevHXki5iv23jpZkud+SYDhsdO+YfHN2TBju+3cauH6JnC4iWCFXjr7ubKcEikLEmYm3eTkiIqXYvFgwxBl3nSdcWV8/NnCN1vnsGRSnlIKGqejcjiuriSXwenFgZhA1DPyHtNJzFyVbOZCPZ3R6RqEnCMxKnXGfofkgk6d6baPmN0wkzpa65MzNU3AGnxuhjxxvFsbTjMQNws/DvnHeNRd8YSrHeP7kADmp/44U5MmRHn82TOmN7z6+MccXh+o44A0CG2jt40h7rKvpXqI6RCse1qK2iCKp+0gg8vDnbOOLBCysl/2uRCY3aAJZStYw49x1dnZ4EJqCFzOff7Errwu205Sp0xcmXD6+MTVaX6Oktj2nToqbXNM9LIER/oOoexe9AaDnAI5+0stLUq1ynpUDichrwdSDsQ8KRrBGBIxbf77vChRAyMu3JSPuf5fdwz/2b9Xew/fyj7T69kWprmLcbm+GJVO14H0Rnlzx4nIqA0bgfNDI+YFG0IKidJg25WYE+3sup9ucN430I20BA5XCyFHlnXhod1z2ToKLAHu3t5zvQaOS8ROg7AINgMiD8tKGErtnYeH4tlnPbjmx5JvSGqjq3LXO1kT6xJYjyttXCjnzvlOERUWVUw6IQM0Z2j/Bu/Il2feweQc+OijTLnbsVZBGusaMYX1GAmpMxjErOSlT7Wi34SKTPuIa14eN3MyvGAsa/IUjtF8y9Yhzyhvj0UagE7+kkeot/b9ccg5U49dhMs8nK7oGycbvoESmfjYPo90xpPt4mnuIt8rnjGIeOfah3dUvS/E9Amn179L1msI19DMKQzWaY/faPhxS5jIZBQVpQ/PDmzz2DTMZ2iYC07HqPRWnKzQBwyh10JgsJcLrjSff8YOjEgt/v3LZaNezljZWKOSOVOHsdfKZT+T80KISg6wxIVcPUaq9+rK8OYb2mFCDLCu0ZX/cTBEMamYdiztkAXJkWHVwwdM0ORKU9NICJ0mgqXKSFDFnwHDb4uJKX+217MtTE8Sy7mq6HS2/cKqlb/6+Z8zSiXGyLk2eswUC65xkosPAhO0ILRh7FulmrKVwaWYv1m+bYSsLIfCPjopr/R94/WVIlZ4eX1iXcXzy3TxOVLz89W+dfZLp1cX1tVqtOKhhL1P0d0Y7PdGlIKGxKslo6HQdzjfduLUS2kQrE+/lEfNuh0mBV9nTyxsXo12arz+JHC5KyCBwymBCEN3NMC6BkSNweZHYPWtXdBZOBFoSt2KR1ZFIR3SUwx3aZVhRpoEA3sKIxPPQOuNPuaDbzrnRTIjhXDIGhMz8njUm/Oi3ic6VnUe5ebweR7W/W0uk1XlP7svPTp9eEZgG4GtBg6HF6Srzxl6orGSp7Hb5lZS5s+wiHp4w3Alt6qSxJ6KXwxhcr+dqOB+vYaNOlf1E3tiBr3Se/EXhc0tbOuoJereKNuO2iC0AmVjlAsfvv4r7vbvaFRonfq2EmPidFoJcSdnX6iElBwVjc+vhikqRggVQnfzeQJU0bh4uGrwxUnU7IuOPmY0l7DoAgxSTLS9U0Z3vhnzSA3fH1WfaXF6toXp+zGr/2fAGGWn9DM/+OEnfPvzr7lZTtgH+O58ZihETW4jiYFBYHtodBNEV/Yu3G676z3EOeF9M9r7ja0mYoBjihzCwr4Mvnt3oW2D65NwuavIdOzfbbe0Yow247wbc+bgsC+PvXYES1wCt+/dM3W4Eo6vVs7lAhHKWdgSnFJixI6mTrPx9JC6otl+Q9HdCbmynLxghJgn09ugw3KIpCy0VpygMKDtPjeptbvNYQjlPOgF1+qooWNwvMkY1UGd6gPZYR2XaAxq8YdaRJ8wwa6F8q2lzPlYq3Vuq2Y3ZFPpbUwUCvTmHceYBmDnObktROaRUsQFohLMh+f6OMPKNFY+/tHvossVIgseheT+PusG6iZviXicufoWMz45cxQ/aE0OtgyHp01PnZkzBmRKCmxAMMVqQ8ZO6ztMOmkSoZWCxIBJp5ZC6Y1lUeKy8lI+5v2bjbv794xNsD1RN+PDu8bNTeXqegVxvMyyJsw6520DPOMuaEdTgNCIi1JHR/OYvKnupnAio3XSOHiqTO109WH6rjvnh32KhH9Tsvz8RQPPtjBFhDo3MSpG6AO5XOjlnhwgSCVK5bTA7f1gUB0vUZ3RFENkbI0+e9et+pAxjJlrqkK3jsx169gNWwbtXOAUGcUz4i/NCFkJSSmlIShJIufi0eHmc07ftgQ4HBwcVjZPmu1j8P79Tr5prC+OpJDZz36cyOvKfhmsOaKCxwDJDCTAE177jOgmQMhGOnhowDAYMhijsqbkK/Dmb1zrYCP4JtGEvvsyv1wqdXOEB+ZaqEtrHJaVrt75SRDUhiurmaRNizDUV/3NI5M8ZrvPqPUJ4APvLB91QI+GWHPNzffeRyHIZFUhEKZfTx2n4p0VLvLsLigcI7LVFdJrlpvPsXBwXjYJxWdQFoQhGTNHHg+bWjQbxBB9oRACgYRTvRSkAU5C4JGtRHfiQXfJgHal9d01YiFi1qjjMlnvQreGiSNteu/sdSPM0NP11UuONwce3tzz4ddnLqVz/1BgJMpWkXedV69PSGykHOh9wXrFJjI4DyAEN/uuAVohJHHDMDgptArbZVDqoNZOsQc3al9lXt5cseQzwtkH+f7R/9+TeZ/Z9WwL09RCovgPGVqDraC1cPfdGyiVde28eHFAWmFvDkQLB6XTGWJEGXRT8qqcm1GGr3SXJSMyePn6JcfTQrXGthc+eXXNp6+PjMs72sNbaEYK2W/yjq95cYFjChkVX9EH9Q2KO9yLq8OPMjU5kZgcIHd3X3j1yZFR4Hy3sRwUi4alTsbTfP3e1zkk7v5wdQF8QJsOStiUchlI1En3jIgatW5Por8xhFHVN0TVM+qkJaw0KA6xsyDYaNy+fSBf6ffbpdGdlDi3aD6Ebi7itIEkpXbfuMGYQlGPDTfDqQX+jzvCdtj3G6/ZJckjg0iMMYkLXrhcDqLqokEbCnag9YWtLLz64nfIp0/oqPOxdCFKQPtwnDAJDZnHIZUHVHiR8+OhhzUwPNbJ48XBRvefKw7XawGmXuQVCFnpY/f8PivYcLuKdTd/W3dNWSnDfXx7wbozvuMoLFcrr4gshzP3Hy483BnbuTFovH//npAc33x1fSACW6wIjbTA4TqRDh7PZGTOfSclL5zWXDNVbCMcBseryDGuqO7oMfDQBmnKSvzTmlqy+fVcx9/PtjA1EboAA9II2NvCuz/5G/rdr9APG7/7WeIQL0S98PqjyJDM3joWoLV9EhATMa4syxXnh8Ld3RnRQFoWBp20dmK+sKzu0aJfqPst78/vWYN5vlcYSFCWGNhL8b8vgmUjp0jrSinVvUo8Bg74pm0vg5QyJp3tvvHhm8HhpOTTynYutPeBZbnCuMdsIyb1h3jaMRCPnmY64E3wlI9UCBXoCs3hYWiGfqTtmx9BeqTuwodzIwxhjRGrg3J2eFlpTrVMFrh0Q3ogHAJYoVnw2Y86BtgV1XP+FDNtuGlUZ15QbeXJ/sHU4XjUkr9ZfGuXMdoc5nePOtcIogQ8J2/MWYhI8tDPbmwVCie2w6d8+jv/FZ//+L/EuCbGjJjTKX3g64r9NKOOJOjsDsSPMNNX9ygEFab9ZnK7sUeMixMQMHHpgbUpfsxYCwwqWAKNjH2nDR8w93GHySDmzPn2gd7PSH0gUKltp/RGipF8ytyETAnG+f5CKQM2fEg/Bm9+eU9KyiF5uAHW0ThIqxFjZN8HKRsvXgnrwZDUWY8rL14EliMcToMknlrc4kaeIRVPZAFzpbn/6/lez7Yw9dluOhIU6n7m7a+/5EV64NWrQHwdOC4BrNMFilWaCaXsxCScrg7TpuEs7/UwEN1B3MclAUydsqgzlrmVDiNwde2K5BgDVdzBLmqk4jeyAeuyour58g9nKHv1xAtz7rZIZ1kShrLXRt873A22h8bhlFELfHhzcb3NK6MzkMVIJ1IvzLUAACAASURBVG/bJfKkknaf01yxq7kupjrobDThshVGawQSrXSsD0brlF0o1UV3us6w0DD9VcNYNZJDoPXC3buNw8h0EUzl6XfPjAEXVVJQ93zZ7JDEnjZ1Kp6ZF6ZwEhwh7Ee97vl2qhgBx8rIdOELAz9y1aETWeK6gpyvOL78lM8++V3S9X9CWH7CsBsCJ0wFNUeAMOOPVPz9r3hhcj2Yxxj16e53jVYnaiDFODVkB8aobrOZibiPc7JlXVyV1Soj4Am53R/qJAGRSO8Vs+r+D/Vuq/WG9UbvhbZvRPMjnpSAjkjMyvXLhVoi5/tCa27GtQGtDC7Ta/nIrIoZTC9ogJ/+zgtyHKxrZzkJpxdH8nGH2AnR00tjSjRxEOE/vOb57RkPvuEZFyaXgvn6GGBdlR99ccNNbCy5Eh+3FsDWqquoEU42WNfM4RB4uN/ZLhs2CsSNq5fVEb2rc4nsceODD6vzovTqzKUlCyKdkZxG2FphMZ2dkSIUUvTY7MUgpkjQwPnBuzVVvj/+9OZarAqX28LhoBzigXcfbvn2qzs+CQunuNAuLh2QFJzfhnvFoqobiNUpmCkHDgfnQXUzxu5RSzSXLNRtR8wVxQmPKfLYocmyD8r19cISkkdOk3jYK5cPDYJSaD5rUnmKxx4zdUQlTuW0i1ORGccU5MkI67MZF2a22ubMiZkIgltUzKite1cqgVILQ7w4mSROVzecPvldlpsfEk6fEw6fg9wQ5JrRMh7jPogpzfCF4DiX4Z9OnzRLDZ6np2PM7LpBl44nkkS8n+qM7up4xOdnvY+nwXwvFQmTTmGFEJQoYW4+Iqk1Ru3YDkM6S16wCue9TE6S/1m1eWSWlUYiuKiTSFzcm5jMzcFmj3CqQcpMboJwdRP56e+94uOPMzHdkw9GOg7ieoZYSGuaaTPzc4vuTXyUjTxpMJ7+9/O9nm1hsqcvR6imBT7+bOWoggQjBnXBWxDyMNKE449mqDYs7OQjPFx8k6LZOK6eShLTIETfUtXmg/DtMiFgc3uUorkuJvjbPFukNUObegMQAiINa4NonoZLH8Tkjv7HOHObMDYxRQa8/2ZnzYlPP72mHCp72Xl431nXI+Oy0ZJDxXJIhOBZaY/IEPe4CcUGIUJKQtsG+xnKg9A3YVRoOywJUlCnQY6BpMjedofV1zkPGh4mKUDSxH7pNOs09QdMH/2EwU3AItDMM83kEZKPvzw8FKEiFnzYjWG1TxTLDM4c5rOvuTDYNmcioYbGA8vxGkzJy4l8/ZKSPkGXTzC9QnUlxgWGkw8kpCd0rTdDAcS9Zm4K7k8Bo8b3+irRgUrEz5gyDcAJxBXtOSVKcVSzG3ndahOikx36iCgDTR0tlXURDusN7fSacrxly2+5aKft36HR9WGPhdyDxPzYWSoMPEn4MZhVgtDN48KMRl4G1zeJFy+PvHh9YjkOlmPhcNVIqZAOlbgOLHQIThSIWUgWoSmEAf3RjvK8C9F/eD3bwvTY7j8idiUKcW2EULEw3K6QvINJS/BYZfPh6uiDUgqjq6tngzzNfUJUDoeVYYN972Txbc3pOs2tFtTq/i5VYPJ+8pKQ4m+xhid4tOaptBJszktcrFjvi+t8REkx0mt7wrqWy+DD251XL65Yl5VtL5zvhh81Vz9qLerJuBYHBCdoIt7B9DamYj2QV7AWOAvc3w7agyN9lewJtqFDAwsDS4YkeTKMmvnQOYXoKF0GW/EOr1cj5OQFdXgOXJxzGRPfQOlkVYU4Ay6ng73pTF4JSm2NgbDvnn3Xmi8jRJS4JIJm8mFFw8rp6hXEA2s+EZYj+XhCDi+Q5SXEdabS+qJBg0yigc+/NApBE6357/jRzOTlfPKQzD2D9lgERBkGOS+0XgEPKCh1gMQpJTAnWEyniWggpZOLN1tHs8/mNBnpRggfdc43X/Plfs96/R3DGr065iXZQGqnb8VxOGXBZpy66UCS0PcCsWPa+Pj1gZsb+NEXVxyvjCH3hNxJ2U8PMXYPslD/PJgBCN0MGmSJhMcMMXl8zf8GGk6ed6F6toVJbWaDWZhrXkEWXBypnjRi3UMHckye9opSDGppWB8ogTVPAaEEYoqoQgg2N26R/hSrBHHJM03DkR6Pg9wm3W/+J//WQHXQS6e16U/C46xrde6RSUIEaulzsKrTB6bcvi+8e3fm9ScHJDxwuRTaLtQHYVkUckDiXAd5wBsmwTsndXGdxsHhGJEunI4rf/9wx+VDI3S4OigEt+GMViB7UUtZSSFQSqHZwNQIcbhlI7kPy1MREyH4EDgE1xS1zTzuSBb387UGAfbNj6m1uw6qW6cOsGhIDDQDDSs5Hglr5Hg4ElMmLYG0+HEqxSNpvcHiFcQT5AOaMpIzIivKwhhKjsk7T1EPNzUfk7gK3V9grnD3o1HvU+ujUyclhmpCJiSvVqdNavCwhhQDrVVSio6NMX9RxOwomTH63LP7MDnFREgrGoLzxcUIurB8/dfU93/HcnpB3S4s+cDYL8Q0sNToe3Oh7l7Q6HKDbgVJg2U1Pv3RkR98fuKwVtbDPXFtDApp9Y7QRDBVmgpoZm913lv+Mg1T4iICOlX5//Cy3/h6noOmZ1uYghn2G01oN3dKuyXJgW6ugzHA8+t7bw6UH0YKyd3bNjyPLT7ORtx468kYTmk0HpnizvTxU4prQjp+nNnL5h2ZGhIGfVR3dA+jC84KlzAFiMroQtmbv51VHWNRGsP8KPjtt3dcvUzkJXH/UHh4KBxvAg/vN9ooXIVITviQV3AG9NQ3qQo5+SC57Z3T9YGXrwbffXVPHJ1DiiSZ+JQYiTnShiNjQwisOVNaQaLQaVPmYMQslN6f1NOtdLbu+qjzZUd0o3U/BndzjEnHsSqIksKChkjKibAupOORsGRCzERdiTET88JyXNFkoB5dlPQI4cDQExZP9JinrSIhZIIuiEVsuKRhDJmhDhOnPL0VbnN5tIy4MvrpfgrfW32Yd1VK3uX6YNyPhZ6wO+b3DJ4u0+axEJ9hJc0EjaRwIMUTaJizyobGA9evP+Lq7hV2CSzL4oPzlKFWqj04g6t50nEydyJ8+oMrxigcryKvfhDJeWddBoejd4XdAnkJ/qIaThMIkh3BOyk6bTfvDBuEUYHqLwx7fIr8v+dv6x/7kf1HvZ5tYRoi8y33gFhG+8JoDdFIFp7eDo9IjqaD1oRhXmD6zART7c4n6o0U0hxqJlChlkobfizRqFO1POitEkIkqmtixmgTWufGYlU/lsQYseaCxbaBmIItjN4nacALVxsGFmiP4LSwcPehsu+B5ZBJ4UJ7SLQSgQr3ndOVQIJwTH78kNkJDN90mTk3ezkNtHzgR791xZtfFu6+7ZRtcEpHhMoYxbGwx0Rvlb1c0AT5IETxzistSh+dJWZUF7bd2C6VfTe2HR72iqbkc7MUCCkQ4wGNgfV4JC4HwuKdQ0oHt9pEQ2MkLy9QOThT/HBCl0w+ZmAH2whiREkEXahdIQZizvS4IBwQWRjmhW/gCwmsgsVp3ZjHskcdEuIiT1W6OLn00Y/nndQgqNGHi2WDuq3Hnd7ulxNmDFQHJXoqjImDClNGSMSwkmVBZZ3fG3q75+uvf8H726/Q1d0HBCN2YVimm6JrIcQdTRfS6KxjsC4LH71WDyXNnXTts6a0OOubMdDmc0/r3Y9s3TtCDLQCGEkig0oQGLUj2ZcefdIvMQ/aEJskV3m+Nt5nW5iwx5repwJZsOZDvpC8aDlI0dfrvfrg1+rw+Y4Je9m8mInPgFpzf12tE88qikxl8eNxru7e3ku0yZP2JArEpgcLWvO1uUpm9Oq+uSKMWsGUVjptcnGGyYyT8iDE1kEHlDF4/93Gj3/7mpsb45tfPXC6V443R+p2oZwVSd3V4Nkg+qzAmXLi2W4aWBZFD53Tzc5PfnrFv/vmHZdz5/WLwOEI27Y7VKy5wXe0Rkwus5CgLmmQCabrlT4a3ZSHrdBGQg8HTlfX5MORuCTIi3dEKRJyQnJG84GhwSO0NaMxempuUESO3vUIxLj4EU3FFdjhgHXXb6ms9BrostIsgUVCyICn4zhTKhDEC5LMF5DMwhSCFyARwaJv+mREhAjqHrihEEJyyoE61tb9uDOTTbyz9RipHd88RsQiYoEcI9ggqpEE1Jo/7MNJEe/vPnD3cI/FhKSVlUazBv1CjkLXhqVBpRLiA4qnB+fUuXmV/BgfoFp32ugMa4gx0oa/WGvbGMNjzGvZ3YmwxCn49LVoDJkcF0bo7uNsg+9jtv7juJ5tYVLUofW4adKaD3JDUH/jdT+ymTrHRrpAGYzNP9Re3B7AHNA6RiRStuYWlD4Q+hOIrF1295e1zhjQpZOSUmnU7hye3u2JVNgatHFh2zyZ1oYhI4A5y2gMF2gO05lM23g84tU+IAhf/fKe6xcrOS/Uek+9wG0pvHid0Z4Z+0ZPnqhKd0642zVca9Pn1ismIR3O3Lw+cvMi8/5rePf+4jaLIUiH0MWPx+Zo27B6zNMwm6bQOYeK0amUOWN6QpaX6HqD5AMhJ58/xUTKEUmRLsqIiaGBoEJKJ5cw0JCoiK7EsNDLhoUwIWgJDRELirFSRyOmK3RZMYvTYBxIcSGQSCkjs0CpZD8aa3qSIfhn40e7GCNDByqBSCSKh3923TwR2XxT9bgoCaKMOaPygboHNdTunW7IR4JkbMpEQNirMaQRzBcRIkLZLm5FiY5ZsaDkZizR6LpxfviOcXngkBtJ70nxQlqUmITlqFy/MJ8zBZd57LVQd9dbrXl1PK8mkgbuH87OeY8RzObMyu8/DZFR4VIaY3Tevr3jfPFj/D8Vpn+EywEg8LhfsTYYlxkumDzJQ8QVutulo5IYFhk77H3GKI/g0eE6kBB9yFwn/2jO/tyfZZTSeMywt+HbnCKVOrVEKXkk99NXha06JE41MRrQnUU0zLc4vQ2fd4jrbGpttAYxLJRm3N82vnuz8eMfv+Lq6pZyKaRkRMvcvrmQSuWkkHN2dMhw6YBpA5lrcnw9H1PncDX44Y8/5t3X79g2pfVIkt3BadE9fl3Mj2JJ0eCdoP8u8XinKQvIx4V8/TE9v2LkF5COdAnk4HO9mCIDqAhdAxIzSkZkcU+ZDkd0WEDIxDhYVFhVWKNiITLCgsVrIDIsYbKgMRBCY0zoXJSESHQbiQVE0/zsvEiEEOfAezwpzjUKYQwijWV0ZOyUccewbXLGldEcj8wYrCnSh9E0o3EBSeS0MliJegIUy8C0t3g+g4EZ+6gEDbRxYdRbVnug1Vvk/JZ292vuvvslVj4Qe+EQjEPqpNjQSQhYVoXYaP3C4ZQZDNp59/+PGNnOje3+zGldaecLOSZ6hZRdzmAilL34cgV/IccR6MVofbCdZ6Ky6zp+YxD+T8Pv/w/X9/W978b5XfM2Ovm8pTePG8ICYybUmjzOdvwNCuJvNR0zJFIY06ulIbJvZQYGQq1Tn4TSqg8OqniiRiudMfwIV4sRQmaJPgvrjwVpAvUfzZ2IOL2ydFrzVTziabNmSq+BN99e+OSTl8SQGPuZGBP7fcP2QRmDNo8decZ+q/YZZOBaoBgD0js5B/Iy+PyLhb/5C+PDu43r6xfEY2W05jSE0tHk8DGCIVHRCEF9nT5Gn3xqnYPyxnJzoMQjstzQLJC0O0OoNVJwbZaE7AKxsSAsGBNaNjEJQYUlCLE+cPfLL/nuck86XvPZb/8zlutP6LZSLVNJWIh+XI8euBTVAXeq2aFxGlyigT51AcC0xEwleW9Ib+R6od1+y1/97F/y5d//nPUYuL6+4ur6irpfePf+O8C4uTmxnq7Q9Qp04ebFD4jLR8T4inT6FFkWOo4UCUGx0ejlwmgVISIxkust8uEv+PWf/REPb/6G9vAGHQ8cj8ZxHRxXncz0hkbfsuYlshwiIRsmnX3fXbwZoM7mziZWyzowjO184XIpqGSmGmJaTtSLe6u0KoweudsK28U3kMJvFqV/+Gw9x+vZFqaZ9Pb0V6MKD+8qOUANzHgfnZYNBQv05ikZQcMsQoM2XByoOguGCBK9Vap9R4Kf12Wum/cZMjgGDqgPjhjxbsePeYajRB45zx6hMw2j5haClBPbXt2awPcqbg3MudeKmXL7fuPNt+95cbNwPj9wvt3pdXD1amVcIg9953S8QlPDdEei649QoyMEvOimmEhrR/K3fPZF4PZ94e6uclhsPqyGqJMot734uhsAxdQFq3ESF9IUG9ZyzyEJrJmqPmvpSZEYoAdUI3HwBLgjPHaXkdqbB8Up2NjJ4x0ffvUXfPj7P4N2pmnk21//W37y+/8Fn37xn4G8RMJrejgiGhDN7h+zhkqcWqTwD/x44IZhmSLKYZ5RF1TRes/+9q/5xb/477j9+hfo5S2yBtoauY3+cy3SCVmxrpy/A9EMFnnflNZXak3k/ILDi9dodvi55/dVbt99S93vkWb06nqltj0QRiFb4xjgtCaSVrJAmlafjksdOkYZzXG3Bin6HT96I6dEqw1rRtBI1AnBNf8zroubr51s4Vo/0eSWJEv0Fri/bdzWQSuRGBKGfV+YnnRNz7c4PdvC5KE93uEYMJowtkjTQLWK6mDYY477mOkZETWlzAF2CB5KSvb2IkSP06nVvWhtzJU1/vKp9TH+yFNTeh+O2zVl2CNjSJ6SO7Aw9SOuU6oTpBaCUyxhhij2eSA1775c+eyiQxnw7a8f+OjlJ+7ir52tN4+YPsLyYuH2TWexyvVrZ/ZoUCyADsPoxOj6qrwO8unM518c+PrvoW4NE0/fGDi+dj/vsPrvtnchJkVkkKKwZkdpONe/cffhDdy85eWLHxDiQrFET/7ZhJh8S2aKdBCJmECy5IGkMrDJNA+hs7//Fd/98mdc8YElN3aDS3ng3//8O/btV3z2W/8N+SpQJQALtTksLSoMzy3yDZq5Z7D1On/vAsG3io8ozF4LOi786R//AW//9t9wknuuDkbOnRgK3ZpHXWUHxQXffbCEDHiXfGkNSQvbpXP5IJim2Q03clLWZLxeIqOd6bXTh7BPmYoQqHuld+UYEqtGV8vnRJXIkEYbu+vzkoJ2Sq+E4EuB1uzpvlnS6lxpG57uwyBOYsP3L2WdY4pG2QqjRC5no2sgpSPL6i9LFXUvJDMB5v//x/r/8fVsC1MXZR2NnUgXaKVQPyycpaMySDnShzBMaDacJkCd+NTuG7GZlluq+7laV4Y4x8gxFUyd8HSkD53HQ0/azcF1UKM9WinUkzy6D5SHied6dZmqY6ZNY+prHh3sAqMbJrhkYBYmM0UanD8MZASOa+LhtiBVKaXDNrs8xCOkpHJ81f3UNCanasZAIdWJnCfleKN88dMTf/eLO/Zy4HQItFZoe0BzJJmjgo1OU4eq9ebAtEXc8jG6Sw2+/eZLXv3k94iHQBnOPnJWUZvHZXXrhXjsFdWNzq1X0ILYRrSd87tvCO2BJTRyhDCUWBqj3fHhL/41t2++44t/3rj+6D+n2w1dDpj6OtuGv1h6r0joIO76V1FkRNoQhg8LyQbCmT/53/4X/u7f/is+isI+IudSuLpJXGmkyRnJENfk3A/1ze0I6gVAHDeCdWJ1f6WOzqCzHCKGw+Ss7KxHkJOHqK6r0UpztlaAy155d9+R5QqlM7gnLG70jTOnUILHgw2Eah7+qRaAMGPhnetUSvFZYoiksFBqpV3avDcbtQV6K8gO2hLITm3C9cdXnK5wxf7Ic5wwEAvz3n+e17MtTN8P5XxQPTpcHipoJzAH3kFpozMwUva/2828HkYYUwKMbd/9jdF9izc0OuRt+DqWMf8584/K0SUuHuz4g9CnI9+6QPfNjbWGmkwtVHj6noY+YUo8r8wLbRehigcRDgv0EoiujEJkQUOidRxVoZ5WO0rg4UNnTEjcvvQnSJV4pAfDZnQ4EJOS1sHnP7zhzVcP3N9XTusVrVZKEI5LJohTEcu4TFi/G1zHGFgMjKmGT1G5bBe2+1sOh89JGn0GhDEkEGwGC8zfdy0FbYPRO80KbT+zaOHhcsu7b74mtILEGYMe1AusQZTAKLd8+fM/4rOfVD76yX/LzgHVw7SQGJgXSkYgaPDtlYHbkoUxKjIGMQR++fM/4//4n/5HPm07jQtYI2ZXcF+2javXGQmdHDMeQ+kmbGVGK5mbkhlCz7jKfGb4PcZW1dZZUkLVY6I0B1qC/VKxYcQkHhBgg9vbW+IC168yMUQkDiQ2j32fqN6U/J5UCVzO+0STBGiuhQs9+LZXlMu5UouxnQf75sieVn0YL83heyIQs/Ly1Us0XmYSMv+ByPT5Xs+3MNkjLcYVt2awb52c5kPbBzp8GKoyB9/DdSfu2xJHRqjL8lX82FWra6Fan8knbcy45UHMkRCUXhyBgbiWqXV/a4kxTadeCHQC5JmFQXB28+PMpWNUU6pBQykdSheaCWMkbCT2WrF3hXcfCmvyWUGt3WdsQbj/dme5WklH4e7NTkiBlJQo/jO3vcL8sznSZXC6Fsal8KPfuuLLX5wxS+R0Q86u57n9cObm45UUF+cg9TERJ/Z9hpkKOQi5d959/RWvPvt9NCaKRPrslIZCHw5JUyfK0WVQx8beztR2RwiD/e1btvv3vAo+dgpBGcOcbKCRVRKl3dPPha//onB18xn59T+njXVuHis2qvOahmK6YJNQ0E2oDvAlq+vQlv3Moe7U23t29WH+2KGJEtdAfWgkBoWNsASXNTx2gtYQZzAwDFT67E69e8Z8KL/ESAweJDpGIyQhRFdq76XR6yDlRO/eVY/u8eOLJmJU4jomsbJ6JmE1gibK1mk7WBuEHhhDCQTKQ6N177Au5+Ki3ubzsa3jtqjhgMA+OnIMHK9PnG6O2Az/fHTLPdam57uTe86FCeFx7eBDQWcr9TbVv8rsNbyGjeaQ+draHJTaNF56kSitu8A3JQw3pu7bPjcfMjPVhqe9Bt/ybMWhaj5o9TyumDN7qX6TTk2LRzQ7FbKO4OB9eFqne2ESLtVTPBpG3Qt970jz2cHbDzufvUosy5EyzkQRWoe+wYetsByORF24rIUYXZUcs5tZm/lsLATPzZPeieuFq5eC6eD2/paXVwfGKCypc7ME3/wMV8WHJCwpotInWM0Ho0kGh9B499WX3P3oa64/v2ZIQkJkCJTW0BgJUailUFuhl8K+nXnYPqBWaDJ49/VbcunEo6IynuZYIeuTpWS0HasbGgZf/vm/4Iv/+hUaOmLfg/r9OKe+dSO6mFADKQWgs4zO5bs3/OyP/pDt3XfkXrGUIBy4u3vAeuXVR26QlkecyNRCNTGQ7t9bcP2b+oasicsRVH02Y4Ongj4a5DVMXZTRaBg+v/IRgrCuvkEL4jjfIN6NWZ2ETHWbTU6Z0DuUwtiNtg+2rZMtMqpQm1F69UI3ZMa/+2fvg/9pocoZyYGPfvQZ68trykye9sse//2MR9/PujABT4/3hOvXMQfWrioevZOWzBjD0ztkomDF297W/c3c+vQ+CYytEnPy4jV8lY0++pGE2oZT6gRaE+pwEZuZEVOidO+CTFxIdynefbXhTvra+/TieXflvEZoJtRuXqQMYCDBxaCalS9/9R3BDrxYZ/Rz747n6IHtofPm3xcgkNZEOXjWm8RZAGeIgYpQekFCQLRxdXPkdBN5eLtjp4W8CBomTygG33t2I4SJCUmejFu7scRAxtNeHsaZen5Lkp/QSDBd/b1Xem8sQbGxO+JFjNZ2GMb+cGFcCrdf3/HanG7QcO9iWjN9FOJkV0ny3yVj5/Lhb/nw1R9z/fl/yuBIKSB6YFjyTlcjWr1rNImEoSxRyb3wB//Df8/f/p//mkPfMAZdhK0Utm1gzbHHn+aF+nBhuQrY5gkqvYAsid6dM5Wj3yNRIeRA3Xx58Pi5l+KbweMh0zbo2oiLzxZ9qq2gRrl01pw5Ho+MttMLeAhNxpp5nlxzoWy1HbEIW6BfOu1itAdjq5VRXQpR+kDF55Q+XmCSMJQGaI7IIbO+PrC+fkFJQleYkE4/CPzG0/Vcr+dbmGS6ax8FlmaUUtmkkoMXoG4AAQ2Zfd9gxoiHkDELXC6VbWuOP2W44zoIEhyfEVRJ0XU9bRi9DoxlIjo6fSiX5i14DIFDXDhvO5dWCcmFbpdteEGaN+1g+HnehMc8u0evVTBxKYMMRJtrqQxyjDSMD7cb18sVNmDNkd6F47pyfrjw8N5ZT8uVIsmLcAwJTQmZw/Btr6h0sirrMdL3wcefHrl9U7xQt0YY5oGMEWJ2NXVrldLdoT4EQkrUWslJuD4kqipf/vWf8epHP0XX7INTFYL6fCPQ6VZQYNvOMAblstPPDbuv7B8K8SqRRMCaq+3HIEaIMlgWuIzICEK7GNR3/N3P/2c+be949fk/Q+UFYyjb7uQCiZVDXohxIQVxYtflzP/+h3/I3/67f0PeK7kb6SncQAjNscWXs4Et5ADaO7U7mXIEY58zqpizc8uHEytDzAxpbGWDGXQq5rTQS+/kJcxgUlgPV2x7861w9G4/5ZVW3e/4cHehjYBEt5GEmaojJtQ6SEFgD9T7itVIL0o9Nw+DGExczZSlKOQgDvJjkFbFUuAinVcfvWAcktOXsd/omP7juJ5tYVIGTaKvRxmY+fHBTOlDqLuvwh91LX1kaiugHbPix74utLHwcN6xJIytE5Kn0AYx928xOB4g5cxAuZRCfYxudm4E+95ZFqVenN28NcOKs5iGMgkAAQlCSJFhgxjUdStjIMQpzOwwHHCnIXnh6p0UfWVvomhK3N0CdZCi0wpiDIgdePPtB64/OjjlUiqHNRJNnrrHPjqm0MYgLxm7Crz+NPPmq5XLvvNCAtYTNEVteNiDGDlHqL5VDDyyqx7Tegc5Nbh/x7d/+wt++PvXdBJmkTwNptIdyl/LoNTiiNgG291O+HDm4d0de4KtKDcnz9ILyVx5bg0NnbR6IV4D1LMf3d796k95/eIlui5seXxNngAAHixJREFUfWO35D48MaxUqrncY7HGl3/yM372B3+Avb91tbwoVrunr7RBNiGijL16Ak4wStkJq7CmRGcQevLRwK6Iug9y2KN5G6IGLg+++fUc0kAv4m6ExRgZgnpa8WBgpbIuaSKbJwHAjHIupNiIKdLEJSWtuqsh6aBeoFyMtncCkX3zF08IrlpnJufkqK5yz0ITxfJATolPf/Bjrj75iJpcJR8w1JQhE/GDPnk/n+v1fAuTDaoeMDkTxyDExOll5noIISZaGzw87Nw9nKnFdUStDzrTJhKUOj1LPfo2bKgfG2w4OiKLuBmzGJdyoSsUMUbym4wAIQpX+YABKQYWTVyZWzIkBgf2R/d+DXM9Ua31CZS27ztC4ny7uWqzDJYYvZMysFAINHIYEAKXsrGuJy5391RthKikGLh/aIwh3H7bub46MFLn/N3O9Ucr3UccLixUf/tCJ66D5Sbw8tMrvvzLt3zMNXEYdb8jXdqMPQdNPudptc+CNGFsquytIiacovDrv/xTPv/xb7O8+JSG0Kyjo8HwAM8YKnlduWydeunYpfP+mze0bafuSrfM3Xl3828AnYGbpm4zSoeARTgihCpslwduv/olr3/7R3QqD+WB43KELmx7pC8ByxW7e8+Xf/wvye8/kOpgWcWDIGloiGj1VXswRQc8fHcmfpQZQX2BYB0NgS7Ced8QMdZDnkSJ7vq3IfQyo7AMrAlBEuuSPe5q36mtEWJjBPOo8tIZRehbJ0l2rVwzfyFJR6NRujHwhB8bsI1BaErf1T2Y4BHlMxFndEganLYqQlgjFroP9q9XPv+934arjxlLcO5Vz2QRggXGnH/5YumfIsL/312e7TMr/CCtQr4aHIKRV0Ml8XIs3H44U3Zj3zoPD/BwHqQVNBk5QaExAhxXhekFM9GpNVHWlEnizCSSYjHAogwVzzu7EjeswtNbT8wNr2FCyQZugzGEUjq1Kq15uoteHCh/YmUPlUpDR3CDMTbRLJ64EsOg1Z01LWw26M11VvVSWPTE8eoV/XLL/dsH8hJpZ+NBKmkIYVF47LzohOTDhOVQuP7kyPs/Wfnyq4UvftC4VtChTzTLYZ4PF6aJt+sgRHF6o+sXSQp3lzve/vorXq0vkJmFJ8E3eSF619j2Qi+VeinU84Xz+1syQt8avQQOJyc2KK5mtt7RJWC4V3FdMmEEbDRaN779+ite/nBHWFkC1O3eww/swKiC9sL25g23X39N2Hak2Aw5dXNuC5WyV5boGjVRuL8ttFZI62A9BMwaGg2b9E0NQrkrxOis86AQY6ZsRtt0iijdVLsXj+eK2ailcP/OzcoaMr1F6qXSd9jKhRjdk5dyxCRgQBntSSAaRFlzxlp3KYv5xi+n5Dwtmc9FDMRjpkplkw7ROL1+zYsfvMKO34P2NCSUKzd9M32gz3mw9BvXsy1M/vvzzdzAWE+BT764Iu0PNHbSsiBE8qvMdq6oZnpX9n2w7xtpUdbrlaYdizCCJ2ekdfGckwkXSzn5ZiRGj1KLypQqgQptcbStmOe10R09kmJk1PoUkz26OiupKVoCYe+UOsgxEqtQ40BjIMVBL+ZHCnMxYorJjwDiQ29LleUQaedOQDmtC+9vCw/7zuuPV9g77V7ZpHG+38hVePXxibiASkWjJ8umoCxXnXxTsXzNn//Vxg8+u8K6UM6uYJbhxQjFHw4VhvimR6OjYrxGd46L8vaXf8cnX/yUJEdK66ivo6ahS1CNHjzaB7Y39g9nTksgjE49V3btHNZMb53QOqThVo+Aq9h1YAprhvNl53z/jru336A3GW3QaqW3Qe+787is8fXPfwH3DxwN1MwH3RHozlfKQbExyDlR+qDu0MsghEFb1RcJQbzTGcayeE5fUwgaKG0wZuClELGQsP+rvXPrlSTL7vpv7UtcMvNcqqqbHjxjLBubkZCQRgjxwDMfha+A+Cy8gISQkEB8AHjjAQPSaARjMfIM4Pa4PZ5L1/2ckxkR+7J4WDsy85yqaoQ8ks/IZ3VXV2dGZMSOHXuvvfa6/P/i2c8zr97e8je+NfLiRYTiWKaMYtbNkpKhrBLIU2Ep5n7InaEaZC1UJ6gzxa4oYYiodyQWctHWNsc4jOY28I4s8K4kZl8YdiPbF1u6Z59RLwZS1xPiFhc6xAW09BasaMzHT3lMf0lRzLLRlpwqoXLxWaQrtoWal5maLRGt6wxpYNN7nne0ujBBSSSsStv1gW7s8SEwl0Ki4oeOpWQqvhEsQmo3FLEIVXTRImQtLyl0kTWuoV0rmcmgxUF1+OBwHcSxwCSwZGoy+BZDsqwgBae+rYzgxYqQ5wP0OwGvbHY9N/OMqtAFwWmyvJ25MJEZxw7vPVkEPOwls7l0dLtI0UquZjW5TtheKf/wH/0B/+Hf/w9u3u3ZXRqzml+EfjS4kiKJrrci1SUZiUJdB7QToleiLLx/+RU3r37O+J1LuthRssOHDtUCMjMMIwd/gKLUJRFRJGW8SqNVt0z82JvVGZxrAGx2b8nmcytF6UMl6MSv/vxLfuu7n6NLoaZELcq0zORpRmrl7c9/gZsyPhUoShRvoU8V+qHDOZiniXlJFDVlg3OQHHN29F2gUlCviItUjRiJeqU48BKZ9wshAg7ezROxs7ytb31xjUjm3etbSzeoFixB1LC3VKgITmOrlQzMBwu+FAz3a7zo6XqHVqNYX2aL7NaGS17UHPOFSirmbgjXGy4+v2R8saHb9bjtJToEpNsQ+l1DJXUg/qiM5DdFK/GYFdORmdU6Vp1CV+ijTfLhwtLySy7EbGHtcVfxXcGLZ9kvVvU/VeMBKx113iPaMQ4DXfTUznKfihjZoA8RVw16t7akTleshkoVqg9olWZSO2I3UIvi8eYHsHoX0pIoRYhSIXimfaLb9mSnULOFpNWTqym8eZ7ZDpFUAvsp0/eFcbsh1wVB6XthHDzTlIk+MO0nbt4phEC3Gch3iducKItjWz3xIsBaYuVgGCrPr77me3+vp94cqEMkuUzZZ6RzDF0gjA4XCiqJcEzTcHQtQtUHcINnSnu++pMf8b2//XfBbVjmSvRGy339ovL1L79uCR7C7bsbOvFEzHoLAZDMkit9daRc6FSIGC51bthZqoIryjg4el+Zbl6i8y2+empaSEth2lfqYeHmMFHuFkjN2gtKFGPd9Y1FZX+YrXpAbVtq5UgeF3xDp8gohktViuF+gWG/++BaOZBhhCOV6JwFTqKnLHdm7VXr7KKFED0+eMAILlK2oIexHFsx8rYfuZv3COClIy2LpaXkxN1dsshwrWgQ7u5m/ADdrqPbbfj882vkYqS7HPFbo3/y/UgYPH034tRKa4Y+kNYKhxWK4DdEHq9iou33W6mIiuA7h+vEEAOl0HkAT9cFcjYuNJyHogwDbRsEpVbUOw5LZr6ZOBzucNsRJyMy9EeWFUUIYhXzhqPscWl1LHl7yeKoWDGwBivLoKqVSBSFYrhJhGBAYkFISfESjIhyVgO3Fw840n7C+cAyJ0QtEW9TrG7Ldz1lTqSccN5Z4XI15+UyFQ57YdgMCIF8mNjXTKkLOx8tVykKFcfYO8rwF/z2typfvZqZ70ZcZ8ktubQ8mpIZBjGUyxnWuhfBHME1KV6F3Qhvbl6yv31H9/wFWgOaPdvxmtv5FS5GcimknDnsJ7btfS5Lxs9KjZb9jHfk2kD0MxAMMcKJMOWEOkfwle0AN+/fc/v2FWyeU5fEfJhbjs+Bw69ec7jZM4on1YwLvm3hrKh7ydkWGm0QMVosAILaxK/ZMK5U6QgWBFgy4rSRGFRUClWsbMl7oescPghQjYUYBwVDKm0ugqILpRiZAa1wG2fIo4rw+tUthUy/67i7zWy2AyXP7A8Tea7kKiQV5lwZn+94/lvX7J7vCJuI347UzuM2HaH3FpkcBlSsPCd6bxZvXVCdGwZ6K6l8xJG4c3m0ismpUkkUOrIkNHhC7xjGhM4F8QYtklIl5RnxZv1oNQpbY66wrGRLphRCMOuoaGWZDoaCOVXohe3FBRo8C4obPNqyaPtqjsnqlNDMaqlqEL4EyxSuLfTujNvM1UApHhciWoRucOhiKIcheiSAFKOgWoqFm/sYcMVB8sz7SrmMqJs4TAtl9gyDt8Q8l3DB4SRy+3pit1HzYdVKcIEyKYd3sO0cGlvtXDwQu4nNridGZZ5mhixEL9QlIwuEEWhAec555qngnPlxShZDuozC7kK4vbnl5Z/+H3736vfBb6jBFoYubHHyliXPuOCIfUe5u6Vo5XAohAE2Fx6HJy0JF8yX51228LtYAqYPnjzbdq8fKuPdwvLua7bjleE/FaHsE8u7menrd4TiWBLEhmS61s3nbCw2zhtFl1m7zpJxUZCKC9HqGXMmumJ2ohNysohs5zxVtCGnWgFs1WT1hWJIq2kx+JFSKj4GylzQxfxbvqGI1ixIFupspBWas2Wc3yppn5j2C9UX1AmLRqTvuP7imvHZju2LLcPFiB88Gh1uGKjBtvDqjPHZe0Oz8CKo7KmSqGVrpTu1M9Pf6qlAyl/FlP7/kkermARACoo5GuelgnZWMOmBBvxW1eHciLhKiObncNIKH10xvO4C+eCMsy0Y75zUwmGZkArz3S11esvu+RUXl1tKV+mkMpdMt1wi0ZPEgNuCCyxzIaAsy56u95Ql4VQoaSbNxXCvxfCcnROokGsmxAyDWjTPGSZ46BTxgTQfEDLihCVlljwTeqVSmBZtJIiOw5QoBZZZcUF4+6s9vsuEvhqAG4J0lXxwhM5TNbLIgo89w+aSOCiHVxPDXQ+xMO489IIsDqKVy5TaYGyNAwCExi4sdNWx7SO//OlP+fbv3DBcf06qkeCFfXaM4yVXl8+Jd5Wfl0CaBDdEVD0lK8uMFZ52nWXzF8EXb4mLIRgcm0SCL9QCnWS6CsvNge2LANmTDoVpyizTzP5uj9/f0auVHHlMuZWltFwjCGp8c9KsGdtSmr/GtQirOKGUiGtQyCF4tMCcKwRl2A0ISikLNLICI+oE52yfmLWQGmCh842TL8qxRjM3PHpFSeot3ysVqjeCi7CLxG3HixfPuHh+SXc1ELaBbhuJvUN8YCngOjmmXKiz1BBDMo3kpIQx4x2UJMwTvHszGYjemvgrj7lKzuTRKiYVATVcbmrh618euLkZeLa9Zty9BlHqsli9lVrF9zAaZVPOppCkMbFWGp9XycTokKoMKkhV3t3eEPDU95lX7+4YLna4cWBzccF2GCk5cUh3bK82eFdIFDRCERuISsFFqMkKSZ13LNOM8wM1V3zoLfLihRiMMluDGjpAEELwLUw+UqbEshRKrRY1k0hOAkmQPkK1MLIj2OTxHTUJ26EjSCZWTznMyOCRgxA2HmJGXKKGjOsD1y8ih5cH0lTpEwQ1uERdMjIIRRPVScubsfIHFeNmc97hc2HsKjd3P+PN1/+T71xekcsVzu8Q35Fa4uHLX77mzdfveJaFMlfcYJjlJVWWWdiqESKUBNUp6tW46HyDxhVLYu20cjX2vCvvqeUGiCgLuQjTBPtDJSYhSseUF6QWqhhOlBdMmSQr83Fe8L5B0NDYU5w5nkStdNdSBhorjlq5jnNKSnc4Z9hfVbF0jVyJEvAChyWBCqUY2WcIEZcjms35PZeJqZYW1BFmChoFCZ7Nsy3X37ri+osLhque7qJHOmdMK65AhOKULljwB+9RowdqoIhmhXnfYYQZMxDQ0vHq64U/++kvGpPLaks+9kq5x6yYGqKkYKUbv3o58/0fvOGzf/xtYklovcO7RL/xjZ22En0lENFaGosIRh3uwbtgSXHOQSMW6PFcjcbN5p1nXhLLq/csvKfGt+QiHGohDB3b3/s2Q3AE7/HAUpXqPSDkWhDnjHa8VIIPpMUcjmlKDZJEiF2HNkVjGe0GZOfUk5cFp4J3CsuB+Ubo+h7JgXQopLDYlrAUUkq4Ci576gSJgAZh9OaoLuJYXGETPT4kpBOCQAwTz55d8io6dK7oAfzGWc3dVKADHyEMhgxaqsEV1VRQD+qVmGGLsvN3/Nn//q88++I7hMvvti0Z5hfaRNK0p6SFeZlJ4nE14rXitOCSYQa5ZBx90sD6vOHEQCloUepS0ASUBc13zNMbnL/GuYIkKAdlvi34xZNyMViWmqERFdScLUetWE2gU3Os51ZcW2uxHDRnBJbijZkE56nZAgDiImVJKBl15m6MnaPzAzVN1NlABJdDQYIRJXgfLTvcwZILd8vMoVQmlDhGthcj46awvdxx9eIZYRvpLiJxG4hjh7redlxGykJd0Q5Q8GLt8JarZZRjgcJMyQd86IxwtUSkbPlv/+VH/PxnbywiJ+43Bvfk8Sqmlr0skqka2M+O//if/oJv//bn/P3vfYfg3gLvUZ3wwVhN7pZK7xz4EdVEdcYLb5aJbXW0VENdTIorFckVDQMlgaij97BCrRQKfQdLmXn55U9JIvQXV4xX14QQ8Z1V6Ke6IM5TKLhoGNwunIG8KVQpVLGSGElCdgZGlnsl1QQhU9bqdirTvKcTg81VVzmkTNz0Vh5RLGt3f0iMEvD+QNHK+5tElcqsleqEMMB27OlcxHml2wQOfeXqxYbbVwfKYlAtDigL1GSlCpoNIZNSUDxtt4CqENXjRNnFwu3dK7788of8nX/wuySdYM5E5+i854svvuAnAjUGshdKMWd6zYmShenO4V1PCYanVVWpC9QgDYpGqdUjweM7Q36Yphu2n71gOghjXFjcgehnRGaIemR66ceOzkXKfkZypUyWT1RKJjXIZKnuZDPUTHCN6snKk6mihKi4HvpxBJfpOk8/dHjnORxmDkmQmui6SKJS8kISIaUD464jjI7bw0S8jDzbXbB7fkW/3VjaSleM+HOI0LXATtv6CwUXjE1YGkqlLdMLsXcsNZNV8QQjaqVlcnuPk8Gs7HLJH33/Z/y7f/OHTAexyPEKyvTIrSV4xIrJcIwxZyOwr54vf5H4F//2h7yZv+C7f7DhcmsJfruLiIRMCYmlVc1XbxTdPgRcM62rKtNhRosB8Rs3V6YKEDli/NRaEQxjaNdfUyvMc6LuEzevJt6+eonrRqS/Yry6ZOgC6rylA6gDb1GuUi0MXWuDAaaQS2YplalY4e9BAiUqWdVIEnygp7J3jroIk3QsrlByzzx3xt7hBCeRVBbuDolnG8du23M73YFrVfcpku+E29eVy7Ajbj11UOSqIteO/buJVCAm6HMkaKTQGe5Qg5CttdL7ntCYSGqtJIzqCZcZO8/rn7/k8OYVw/UFyjM+2/bc/WJht/1twvAFaZmZO2XpF/ZOSQL4DipodeCsFES1ItWmYGn0SkWE6oTsbavnQ2QzDOSNJ28D03Bg3F4TY6UXh5LxQ4DG2rK5dNQ5s+wnSqmkeWaeC7RSnFJWBMdKCELprSD38y8u2e48iOFqK2KsMJptSy2BkHbc/ewl+VDptGfpRooD6QPdpuPisyvGy47PYiAOHa7riH2P78yqCr7HdcEA6XyDP14M8tmFBNVRqcex6A22lJydpTGIQDVl7yQSfCQnj+qO6Tbxg+//hH/9L/8zP/2TW7RabpsKjxxQ9ySPVzE1URo9OAOlCj/5KvHP/9Uv+L2/1fH7v3PJ9cWOYTSCxSyJuMLOOnsFVvzocRiJYsrDEcMGNezuWlPL1Tf65JVqWlSJzYlZ5kDJkWWGUj2pCCW/p3a3uN2AdgYHEsScprW2qFsuDfnS0AtqKaSUWabKPGVKg+0t1ZszHE9XHb0qnSo1BUp2ZAfh9ULfezyOsixHeODhTWTshGWJZspHR/+Vw4+KjJ7xWU/cekQK5X2mvt7y9Z/bvXd/2hE3HjeIRSOjkTW42Bt4mjdSyRXIPjlTGl46UhZupzt+/IP/zuXnb5g1o7Lw9tVLfvzDP+blV3s2TtkEx0XcMW7B+dIKhx3DRmzbGB0uGqSvW6m1qoGlaarUCvsCuTtwcf0KTYH3N7fcvlmYX4/EbDwg+GrML1qIzhFUKLPgGA0PexHmJbVIlSng9WY+OC62PRfXPa+rI3RWYHyYzF/jfaTrNhQtlKSNMefbzHd3Nta6SI0O6R1y54i3xvYsosBs21WfGkmn0EXf4HAsCkl15ogXIYWlERbUYypDbOPVkhTUtnMY8SfV6LjmufD2zcz/+vGX/OhHX/HudUbzFRYSWZrT+3xmPV55tIpJMbojW7EquImCWt7Nm8jLt5nv/9ErvLfqbxvJFi49XqNNJllzoRpTnWuKZ303/gH2sZ3fypJo8QulsaCcXIdFDE5iheABOSJBrucev6cVf3Jif1nvQTtP28DxClEtD8upnVjOUHTcqkDFNp1Sb49khoqzjGGEIpYlLKKWl0PFq5EYSHVWxMttq523NojY7+oxXZh7Y9iCEutjaWvRl1aISmpU5g31ofW/w8iElWZRquVwIQ6VSjWOa+sGOf1PhydUR6mVIvZH5Y9bQ1vhV12r5PXYp75KK2YCtCFTthsc3+X6VtYvRI7+F1WM8quNIT3SVdwXrcYNdxyBclIainBEYW2sJHLGTuL0NA6OGdnti7IOlge3tLEmHzalPZxAA1QUVB1oREgoE96V9QxQo796zPJoFZMH0BNz6BrutE41mp6KHle944p+/tKOI51GrX2qwLPDNqjNE/Tx1SSdfy9tLhgq9pmyuC8Pvz33N4p+PCIiZwOuaiOSPIYAlLw+DmeWna7eMFp/mFPdoDqsqHh9NndU3w3fioqK+XbsPtKuJ+2KnzL5bUDrUSlV24atVO5wzJOpnJA8s9JYboWV5orW4mMHrZp6ZUJh7W97Fhp3XNH7eThCowZvY+D8vciqhT76bk7HVJuCk9ViWxXH2t9nCx4nZShZz6/W3szp83pTU3DleMa5Wjgfe+s19N6V2jG5Tx9wzhMnchrbotKuWYEFfGk9+FQr95cWOft7fUnS0K5W6NdmxrAOHPPtnaAc1qF1WlE/lpp/XP4/cne7wuozPCrJ1SKQB1fSs8Fx9n09G0DuaLc9fErLqTn/XI9PqSROq7yIHq3yVemc1MT5VNKzK5w9T1NYWc3iU3EN0t8iY14/nR+s69TQ9XrrPzQlr8cuPQ9OF3GnPhQFTspl7Q9dGV+aFK3Hd61Y2VE9u9d5HrP9zCZ0OZvqD5/kU3aCgm29zhYIvafUjm//tECIHBeG4/s4ty6PI/f+YnT+qa21Z7JaWh8uXtav50pX7h0zItDTHUwJVbJq8+MJonIcz49ZHq1iWvyZtdS6e/3vapSeKI/XleD+tF9X4PVlyQPzeK245tyKEeF8BVuVoZ4NVlWhyol08ST/7+XoodI6n0BiWuLe/Y5W2j2r6+x5AJF6PNkU6TppTxNd2tbHfmuMxSrO0BDP2uYQXOV47YeiD7p3VT96bM+HWw1bEjxHg+TY30dVezRe1t5fAQvQE/qiStuiYEQSq4UjZ/c5nvdBC8777vjpE+attn/PlYp+cN5xidPVUrn/d1m3iedW6NqANhbPr3FsE/6koM/HmD5o7r1D0t57s5TUaKJsKXF4ta2za7euWls073HKo1VM/+Sf/dPThwdjoj4cTE0EObM67ouix+2HnftpOd/zu3sr4Nnf8g2T9xuurQ+ss/vtuL8NOL+fnFmGJ1uhbSeO9tCHU+i8O+7ZhHo6flQR68T6oHM+pkrb7/R0XOQjnfXBMz64hrBuAD84eW3P6Zk+XDA+9sNf/27lI4qpSZX79z1+0k+PU7CSuvvW3vnt7v/wtHA81Exn5+hpPBg65andpphaf3Ydn//Nb5n6aoGhxyiin5hcf9WSazqOxfsbnG+a+PKRAdt+I58eXPrB92eDrZ5e8cnEX8+qn5wEn4KY+NBzcH7sfgvPBzmrVaRnqqk1qsqn23GuZR6+6wahdJwHcvbdJy/3QQtP0/Go+M/f2dGkuK/QHl711MaPNv0k5+bdJy/46y1V1Qftf3ive+eeK6pvmFofjrnjgftXl4cP+qmLrn5B/eDcSiMhqOYGqd7wooIIwaqRH508WotJpTWtddvHnYX3+9Qc0h+Xo5N4vf4Hxv75dvHsmu5jr9rE4z85TL5p+HzKgK6tXQ8VMZwcumfxprPf+Y+2vW0KPjmXlZMf6+Fvv2kafjiSV2+Wnk5idcrb59zc8B+zEd0nlNFHbaJ1p+VO93v45tZn+nXJJ5UIHIMGwNl2v7X3IzvFVe69iw90/bllf25Ff+OSfK8fzq+ZRfEI0cPqvfJYMfpjlUdrMT3JkzzJX195vN6vJ3mSJ/lrK0+K6Ume5EkenTwppid5kid5dPKkmJ7kSZ7k0cmTYnqSJ3mSRydPiulJnuRJHp38Xzx1DJzgfxB7AAAAAElFTkSuQmCC\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAT0AAAD3CAYAAABvn4P7AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOy9e7RvSVXf+5lVtdb6Pfd7n3c/oHnIS/CJPEyIGB8kKhoRFeI7qInR5OqNxjGiJterjlwTNfdGzUgcGsQH5KImQlQiAgoKCNI8G2j69OnT55x9zn7+3r/fWquq5v2j1m42R5o+jXBab+85xh7jt361flW1qmZ9a87vnLW2qCrHcizHciyPFjGPdAeO5ViO5VhuphyD3rEcy7E8quQY9I7lWI7lUSXHoHcsx3Isjyo5Br1jOZZjeVTJMegdy7Ecy6NKjkHvBkVEnicilx7B9n9FRH78JrTzEhF53ae7nb+qiMgTReROERmLyPc+0v05FBH5MRF5xSPYvorI425CO7c3bblPd1ufarlh0BORN4rIgYgUn84OfbrkZoHG33RR1V9T1S95pPtxA/IvgDeoal9V/8Mj3Zlj+ZsjNwR6InI78IWAAl/5aezP/y/kb+Lu9zdQbgPe/2CFImJvYl+O5a8gn+r18pD1qepD/gE/ArwF+PfAa64reyPwHUeuvwV485HrLwE+BAyBnwfedHh/c+9bgJ8BBsB54NnN9/cD28A3H6mrAH4auAhcA34RaDdlzwMuAd/f/G4L+Nam7GVADVTABPjd5vszwKuBHeBe4HuPtNUGfgU4AD4A/O/ApU8wRgr8E+Bu4N7mu59rnmMEvBP4wiP3/xjwKuDlwJi0gD/3SPlnAX/RlL0S+E3gx4+U/yPgI8A+8D+AM9f15R83fRkD/wdwB/CnTV9eBeQP8hzXz98N1wWsAq9pxvOg+XzuSF2PAf64qecPgf8IvOJI+Rc09Q6AdwPPe5A+/hEQgEUzn09o5uoXgP8JTIEvBp5E0s9BM75feaSOXyHp4+81dbwFOAX8bNP3DwKf9Qnm+ynA/2rG/xrww0fm9egz/TfgKkn//xh4ypGyF5B0awxcBn6g+X6jGbtBU/+fAOYG16oC39XM16AZY2nK7mjGbg/YBX4NWDny2wskPX9PM4a/BJxsxuhwzlabe29v2noZcIW03n7guvE9qq/P48j6adr6waatEnDANwH3Nf37V809X9zcb4AfAu5pyl8FrF3Xl28nYcMff8IxusGB/AhJ8T+HBB4nbwT0mskbAV/TPNT3Nb8/Cnoe+FbAAj/edPo/kgDuS5rB7jX3/wxpga8BfeB3gZ88Mqge+DdARlKo2ZFJun4SDAmIfgTIgceSQPdLm/KfIinbGnAL8D4eGvT+V3P/IRC/FFhvnv37ScrfOrI4Fk0/LfCTwFubsryZ/H/ePMvXNuP24035F5GU9rObcfq/j05005f/DiyRFmcJvL55xmXSQvvmB3mOB+bv4dbVPOs/ADrN/Pw34HeO1PVnpE0rB57b6MYrmrKzJGV+QTM3f7e53nyQfr6Rj9W7XyEBy3Oa3/dJevvDTXtfRNKlJx65f5ek0y0SGNxLWniHuviGB2m7T1rk39/8tg8880FA79ua8oIEqHceKdui2QhJG8ZnN59/krShZ83fF/JR4Pp54OcfQg9fA6wAt5I2oC9ryh7XjGsBbJJA+GevA6K3koDuLMl4+AvSBnw4Rj96HdD8BtAFnta09cVHxvehQO9O0tpqA08mbT7Pbebrp0k6f1jf9zV9O9f0/z8Bv3FdX17e9KX9VwK9phM1sNFcfxD45zcIet8E/NmRMiFZPkdB7+4j5U9rOn8UVPeAZzS/nQJ3HCl7Fh+1qp4HzAF3pHwb+IIHmYRnAheve9Z/Cfxy8/n8obI01y/joUHvix5iLA+Apx9ZHH94pOzJwLz5/LdIu6ccKf9TPgp6vwT82yNlvWaObj/Sl+ccKX8n8INHrv8dR5T9BkDvk63rGcBB8/lW0qbUOVL+Cj4Kej8I/Op1v/8DHhycr9e7XwFefuT6C0mbjDny3W8AP3bk/v98pOyfAnddp4uDB2n7G4B3PUjZj3EE9K4rW2nGc7m5vgh8J7B03X3/hrTRPO6h1ueD6OFzj1y/CvihB7n3hUefgwRELzly/WrgF64bo99pPt/etPUZR8r/LfBLD7LensdfBr1vO3L9IzQg1lx3SJ7ZIejdBTz/SPlpks67I3157I2M0Y1wet8MvE5Vd5vrX2++uxE5QwI5ADT19voI6LUjn+fNfdd/1yPtTB3gnSIyEJEB8PvN94eyp6r+yPWs+e3Hk9uAM4d1NfX9MGmX+0t9J1leDyVH70dEfkBE7hKRYVP/Msn6PZSr1/W11fARZ4DLzXh9vPbPHL1W1Qlpczh75J7rx/DjjemNyg3VJSIdEflPInKfiIxIlsRKw6+dAfZVdXbkt0fH6zbgRdfNx3NJyn2jcrS+M8D9qhqPfHcfn5oxuoXkZn1CERErIj8lIvc043GhKTrUgX9AsmzvE5E3icizmu//L5KV+joROS8iP/RQbV0n1+vV4fycFJHfFJHLTX9ewcfqIzz8Mbl+jZx5GP38S/N1eNHoyd6R8tuA3z6iG3eRKI6TR+75mPX3YPIJQU9E2sDXAX9bRK6KyFWSy/V0EXl6c9uUBEaHcurI5y2SOXpYnxy9fpiySxr0p6jqSvO3rKo3unj1uuv7SVbiypG/vqq+4Ejfbzly/60Ppw0R+UJShPHrSC72Csn9khuoZws424zXx2v/CkkJDtvqklzLyzdQ96dTvh94IsnVWyJZrJCeeQtYE5GjunJ0fO8nWXpH56Orqj/1MNo/OsdXgFtE5KiO38qnZozuJ7n3DyXfCHwViV9cJlkk0OiAqv65qn4VcAL4HZJVhqqOVfX7VfWxpMDh/yYiz/8U9PsnSGP0tGZ+XsqN6eMnkuvXyJXm8yfChUM5Ol/XY0WbpNOHcj/w5dfpR0tVj87n9Wv848pDWXovJKHpk0muyjNI5PCfkFxXSH751zS7/ONIZOKhvBZ4moi8sLFg/gkf/+EfUpod+z8DPyMiJwBE5KyIfOkNVnGNj1XUtwNjEflBEWk3u/JTReTzmvJXAf9SRFZF5BzJtH840ie5czuAE5EfIfFiNyJ/1vz2e0UkE5GvAT7/SPlvAN8qIs9oUoh+Anibql54mH38VEuftDENRGQN+NHDAlW9D3gH8GMikjdWzVcc+e0rgK8QkS9t5qLV5EZ+spvk20hWzr9oxvB5TXu/+UnWd1ReA5wWkX8mIoWI9EXkmR/nvj6JA90jAcBPHBY0Y/ASEVlW1ZrEb8am7O+LyOOaTW9IWoPxL9X+8KVP4s2GInKWFLT4q8q/atb+U0jc/Cub7+8EXiAiayJyCvhnD1HP/0ua/2eLSE6iCY4C8i8C/6eI3AYgIpsi8lWfTIcfCvS+mcRxXVTVq4d/wP8DvKQBsp8h+d7XgP9KiggB0LjELyL5+nsk8HwHSRE+GflBktn/1sY8/0OSZXEj8kvAkxvz+HdUNQB/nwTk95Isyf9C2pEB/jXJXL8XeB3wqw+zr39Acr8/3NSz4AbNb1WtSMGfbyFF714M/NaR8j8kRbdeTdoh7wC+/mH279MhP0sipXdJpPPvX1f+EhIPu0cKFLySRhdU9X6SVfTDpI3iftKi/KQS6Jsx/Argy5v+/DzwTar6wU+mvuvqHpMCAl9BciXvBv7Ox7n15aS5v0wK+Lz1uvJ/CFxodPm7SOMD8HiSbk9IG+DPq+obAETkF0XkFz/Jrv9rUvBrSDJIfusT335D8ibSmnw98NOqepjY/qukCPwF0vp55cf9dSOq+n6SYfGbJJ2ekDj5Q6z4OVIQ83UiMiaN5cfbaB5SDiNCN0UaV+MSiSx9w01r+Fj+WoqIvBL4oKr+6EPefCyPKhGRHinl5vGqeu+nsu5P+zG0xl1ZadywHyaZrNfveMfyKBAR+TwRuUNEjIh8Gcmy+51Hul/H8tdDROQrGle5S0pZeS8fDf58yuRmnL19FinStUtyB16oqvOb0O6x/PWTU6RUkwnwH4DvVtV3PaI9Opa/TvJVpEDIFZKL//X6aXBFb6p7eyzHcizH8kjL8VtWjuVYjuVRJTf9YPyfvP5H1QcFhVhD8EpZ1/gYiARQhRiRqFgPWillOcKHOc4qzkAmNaKTVKH0qDXDR6iDYG2LIl/C5JaQKWoEjAERDBYrhiLPcVYwGWAgs4JqDRbEOMRY0EinXVAUBd6DMZa6rilalqXlFle3LzIZD3CZJXMZLdcm0z6v/OU38tu//lZmQwWpEFMRAFWDEDECRAdktHrwZV/9Obz0O17A5plVEAvGgRhUIAqAEEUQY4kREEEQUAtqEKOIRIwBjQaNDmsLnM1IRrwBBGMcInLkKA7EGEGUlN4UiTGgKERHXXuG0yGj6RjvleWlVU6f3KRt5sTJZS68902cf8ebMHs79E1GnrXwITKbzwgx4H0keEOR92m321T1HO9r8jwnasRZh3MZIhxm2ON9TVXXBI2ocYgx1L5GNKCxxteR4COtwmElYB1IboiZIV9eJl/uo36EVCOCCvMSltc3efJTn8TKRofBaJsL91ygnkyp5ws6RZu1tZPUwSGFY/PcGq4T2T+4Rn0wYT4YsrzUBmqqcoGxOdO5UkWDt4alEydZP3OS4AOTwZDR/oxrl+HtfzrgbW874OqeMg8C0ZFpwFFjUKJAjRCwWBHaEskMuMxg2zmf8Yyn8rLv+y5O3doFucjg4D2M9u8hzPYIswOML2kXBZbIYj4jlJ4ia2MdWFthbERMoGhZrBVirDHWNLoHiCJOEAMxBgBUHOJ6uO46PsJ8XjOezvE+UC9KcluzspxhqKmqGhFwLkMlY39QEymweUaWOXqdFp12hpVIPd2DMEU0IAIiFo2gXkEFDNzxteVfNVfwYclNBz2tJ6gPgCIRCAGjEaOSFjUWg8GoxaiA8ThriV6QoKAlWeEJfgKqOJdRVxFigVEhMxZnTXq0qESjRA0QA8aAQSEYMBb1zXJ3joigPsNmGcZmWAdZXqBR8T6gMYBRiqJFVdVUpQdxoAbEkbc7bKys8dXf+Dwm4zl/8N/fzWKcoVExpiYSUQxRFSMeVCknjte/9k6KVsE3futXsn6iBwJeFAWMJLBGAVUMBlWDNRZjXTNeoCSFinjECCIBxZEAL4He9UdxrLUJPBuwQ5QonnoRmU6mTKdTxARW+3163SWKIiPWI/ZnO4SDC5Q7F7HzCUYMIUTmfo4g5M4RoiBEfOUZDUcs5jV5bsEYytLjnAUr1HWFqqYFlDmy3OEyh4pBbIEilOUCFY/GkqqcU1dK7StsYcnaBTGzbJw5zS1PfAKt5S7bW/dz+SMfwCJYFzG2QMmYl4FOf52llQMuHexxcmOFc2fP4bIWFy5dZjar6Ps2a/kq3WyJq4t9XAFeF4AQVKjKgIglLzI6nQ5OImFeYl2LbusUF/eH/P7vf5g77xyxP4zUGtOzISQNg9jUpWLIUJZawi0nuvR7jul8wjyUjA8+wAff/WtoWKX29+HYoh0maKxw7UjmHMqMuqowtqZWqCqH8UKro+SZYJ0FCWCEdrsgxkhdVUQEaw1qJIGgCFYMxhZU6ugtn6G9dIrFbMK1K1fYubZF4WpW+m2cqdEIRZZg41A9V5faDEYRQs7qiTOcOHWaot1menCJwWRIkbUIfo6iqAjRpGUTQ8Q8AvTaTQe9EK6ivkQxEARRcAJiDOlsdQuNOcQMEYMPipGMPMuBiswZskKxMkeJ2GKVDIHaYEyOMTk+KM4JaETwWCkRWeCMx0nAKGidjCUIBClQOmB6hFBgbIc8zxERvC/RCFFrWnlGXjgODnbxPmLEoSpkrqC/1Kf0FasnunzDt/1dvA/80Wvfx3ySQVSM8URNQ55AKqAqjA4ir/mtt2Jdmxe/9MtZP9nBuKQcENEIooKxyfpTDMbYJm0zopqsONWIMbYB9oiggCStRDg83HFoVcUYUY0oUPtAWVXMpyXlPJDZgo3VU7S7ligLqnLEYHdMCHNcOUS37iPu79MJAQ01eEXFElUAxTpDNJr6YiLz2YRyYRFraLVaOOuoK00WhyjGQllOQTxCjrVdbPME1mT4EFG15K0Oee5xBlqZw2Y5JUJ/eQ3nLHtblwjzCY9/whNB4Py993FwsMN952H9xBKupewfXKXdtayfXGJeTxjsXmE0H7F2Yp3FYsA9H7rKYjrFmJp57VHbYbm/iSsi+3u7LC31qX2gqjytluCiY/uq533vHfB7r73Ae947YFYKSgQUEx2ihoiiZAQEJJI7ZXMl4zMes8xjb2uzvFQTpEd7qU1nuWJt8wLzwYcwOgJqiszRyh3R1tSxImokbwvOZbQKw2RYUXulYwvERkxmUI0EjZRl2Vj3aQO0JsOrR60hhkj0ivGe6XzKeHaetVMOJ238IrK21KKTW2IYE0Kd7o9puzQmIlZpZ5buyT7TMsPPI7MDGO7ucLB9H9bPsC0lqCLOINagXnHWUoUa1ZvPsN100IvxHsSBJUOsQzSjDjGhPjm1zyD2cXYJowVZFiEqUhiWV1ZZWs5YzK8QqyVAMVmLE+2zTMeRg4MZGhWMIqYGU1HHETAmcxXGVFgj5M6iUqPGE7UmKoS4hnAS5xxZ3qNo5dRhQe09MVYYE8iLnNrPmc5GxBgwYrE2Y2lplSwrmM/miDVsnOnzDd/2xWiMvPH3389s3ACfhAaOHODTslZhfLDgt1/1R4Dw4n/4AjZO9QhSIwbQBChIAgmTzD5iBBFpXAYBkvsqKCb5jCAheTPiHgC7Q0svxkhVe+bzmsVijq89nVbByc0VnMmoqgX7e7uUYQAyg3JBnA7Jp0PqrS2q/SGuqokxECTV6QNoDAQ1RCygWAtSGGKEECOzWXqLULvdSoBnIkqZ3DAVNHp8XaExA2OofEUVKoyJFLkhz3NabYcJhslkwcxEtne22B1t4Yzy2Ntvo2i3uba3jSssK0UPIxVajwl4OgV01pbBKoO9fcbTKVmmrCzl7O0fUNfK2sYyVTkjX90gL/oIOdPpAWvnNmi5nKtX95CsIMgS91yAd/7FiNf9wQXO3zulrA1CSDa2OgRDlEAwgagZIkonj9x+ts3jbu9w5iTccUfO6TNL2KwGE5mVExbVFBtJOiIBbMSTQAajZEYoMoORDIkZvb4yHkwxtsYYi2oEEWJUxAiowRpHCIEYFY0QUIIHp5KswFJYTA84v30nszKCmXP2TAFWseITjKvgPSCQm6R/VmrUj6nncHnrKlYv0GkJrWxBy5XUgGQGXyfNV6+oFTQ2FulNlpsOetbOCWGC2AwjXYI3BK2pfUXla3zIIKyS5acx0gOjOKcULej1Ink7kHeW8XUXVchyRyIJIqFWFosFPpZghMCUqtoCc0CMFeoyTJ4TJcPaSGSGxhJr+1hTgi3BLGi1AUl98sEjGrEOWm3HZDKiqkqkAbx+f4V2q8diXoGmHV6NcurWNb7uW55PVOWNv/c+FlNHstMiaFJmaWgNjYbpQcnvvvqNZHnBi176AlY3C0Q8kYAaRWNALKh6BMGII0Qw5qPAJ1jQgGpExKNiUQ0J8KJJyq5KXddMpjOm0xHOOdaXl2kVXaJOKOf3s3cwSW5krKnDlGpywPDiRYbXztPTMSta4+YTiIGooI0laQ1EsZR14olEDNYJooJVi41CWUamkwW+Vnq9dgJhgSzLcDbH1zViFI0VqCX6CmLAGUsWhcI6qsrjxbN8yya3nlqju9Zle+sy8+GIi+cvMl2UYAN3POE2VjaWONjb4tqlC2ysL3Fmc52828H7SBn3iRI5e2ad1aU+RIN1GXVdsbs9Zm29hc3A5J5u2xJ9YFFD0V9nUXZ4+zvmvP6PPsxdH5xxcOAJUTAELGDVEoEgniiJh3DG02/Dbeccn/mUgrNn4cy5No99/DrttjAajpiMFyzmE4IvMSan0xWKQrAuoDZiM0vLCEaELEv2cKhrVGKy8KxBREE91lrarSIBXUj8naoSfCDEgMVCsimwmSOLkRgrip6l1Q5EiYgsCAqCUEVFjUOtNjQMxBAwNhCjxxhYXQGnNbk1aKjJC4t1QogRBNRHgg9I4xXEj3kfxM2Rm2/phRojKWBR1YGyUkKMBK3xYYpGi4aS2XRB262y3FlhqdcjhJKqLhnvlSgWYzsJZHSEqqdlctqFpdcuGE2nDKcD5v6AaAaIjPAhYEwH8VlDsAsuCxiJDdFbYcwEY1uILZlVcyofQMGK0u2l9xpMp7OGjDXkeYuV5RU0xAeOOsfoQRQ1cO6xJ3nJt38ZIQTe9Afvo5ol60doDDdiAj4MGgzD3Rm//co/RIzlxd/0AlbWc5AKK4I2VK+KQhO4SO6rwYg2dSUO6bAfESViUK0IXpjPSkajCXUVKQrL6soy/U6BxIrJwUUORnezWMwx2gPtMZnUbN13H5fvfh/14DKnl5Tlkw7n53gp8VEhWgxpERojiR9UJarB2QzE4H0gBEXEYgRiFKaTKSF4ut0cl1mIFjAYscQYsDZgTLIYjY/kCA6bgNAonfU+Z570GMRFhgd77F3Zxs89WdZFnSC5Mh6PWV3v024XqAh1pUwHFcNhyczPqSlpLWdgLVev7oAKS8vCws9Y7udkzhPDiOmoZFEpK8unCdJhZ7fmj990hTe/+SqXr9SUdQoEJbATUsgqEiQFLQRLYeDkmuEJd+R89ucs88Qn9lhaE/rrXaJUXLo65mB3wmI6ZaWfsbxm6PUU5ypEFOeKxAqLYg83OqNo9ASFOkS8RjKTpaCdEYpcMEaJUXHOEoISg282SMGIwWaWoEoUQ2YdJlMUj4QamylZDhioo1CroHjUJqDSaBARnIIxSl5AXxStSzKbIeJwmSFIQNRQGCGG0LCbkDyBmxrDAB6JQIZ4BEuMSggVMUZiEKKHUBli8MQwwNclamacWG2RZw5allmomIwHzBcBa5dBhOAHtApDsbJOu1ugvqRVlFzd3WLhB6ibY0LEWIuKJrLfKj4aTDQgNk2kGYGJRPWMJjle+xgc1jha7Rb9pT6j4QGLRU0MQp61OLFxAmcts9kU6xI3YYxJO1qMYOHcHSf5ju95IWB4w+veTTWLGLUJ+AjIA+fIDSEYBtsTXv3rv4dzhhe/9MvpreXJTdVI4uYO6/cYSVauTb5uY0FGFEmRXFGqumYwHHFwMAR1rCyvsb6+ROYs5XzOtSuXKBcXONh/L4v5VYysYfQcw70JH3zv3Vy9eBd9N+aWDWFjxdFqLXAmEhRCbSA4NBqCTxxh5SM+AsZQq2IdKSKM4kNFVBBjyTNLVS1QapaXe4RgsCQawNcVdV1jrSciZNZiSG7arK5orffpra9ybecao8EuUnnapsvq2dOsndjAFgHPjCtXLnDXu+8iyw3G9tjenZPbmlbfsXK6R391k8Fgj+Fwh1ApuXO4vKTIMrpryxiXYbIWcdYicxmD/cC7/mKLt/zxgDvvHDCeKVVoNiFSQACUZJ+btDmj9NvKZzyuzed/7gpPeWqL02cFm5XYjqWmZHwQOf+RS3SccHqjYKnnyYuAyT1RA8Zkib9uMgh8HWi1c6xNrnNeGHwN0dMEKTw2M4gRoqbAkbOOxSJRJsYIYNOmIkrwSU9Qg5VkKVqJGKNoVGJIQbVOkafgUkjeRmYOI8JV400kQFZnQJKO1FUNEikyQxSDGINtUEcM2Efgpf43/385GEtVAUSCVgQfCbXD155QKSKOWCvlbIJmwqIcJQIVxyIsmM9nDMYLymqKAq3cs9QrsFhKV2HU4+OcqpowW0woOhZjMkIM1AQkGsgabiQ4wOIyg3OKmjngqCaXiOYkNl/BGUOnXVBXJbu7e9RVIM9brG+s0u11WMxLnMtBGisPmqioIGIRsZy59RTf8p0voqrhzW94N/UsgiZ39BD4LIpgCEEY7kx51a++Fpc5XviiL6K36g6DsCBNmklj0hmTcltcE7PwdaT2kcm8YmdvxGA8RYxjY+MEG+sbWBuYzQdsbQ0oZwMMW8R4N3W4F2cd8+GEd731vXz4/bvMx1c5ua70ThmKdsC2lXwlY22lw/7Wgvm8IrNt/Kwm+ghRiD5ZnxoFcUIIPi0u9SihIfhTWpAT8L5iMpnR6fSxkiEYjBhiFGLwgEfVE4hEG/CZYXl9ldtvu43ZbMT42hZhuqCqArPZRablPp2uZVGO2d/fp9Vr85jHP43Oygr3nb/A1UsXWen36PX6WBNZXV1ipd/CqFDOK+azElt0WFRNBlFu8aWwt294xzsH/Mmbd/nQXQtmcwhEkIjIYZ9jekYxqFraBdx6Nuc5zz7JU55ScPYcnDpdUNczdnenlIMIWjPYnrLWMtx6ukO7NcfaEjJQa8mMS1SNV+bzEu8DeW6oqooszzFWiBIpOoaoEY0VmXMYIASPiKLq8ZICKDEGojFIE2wCSda1j8kbkkiWpYCDGAjRN9H+CLHGa5a8ISD4CjE+UUtBiSoNnwjeR7yPRFUyJ0iegm9qkzcSfSC98+NRwOk5u4aXOd6XhHAYrg6Awdqc4BVVWFtbo9vqM56MEFmgklP6yP5wznBcJYUDxtRUcyUuOxauxFCjWrK2tk5RFoynQ2IAl1nkMIVDHSEISAvncpxrY8XiQ5UUpJpTVdu0eg5btHDGsLu7Q1VV5HnB2toqKyurTGcLQh2x1iHNTifGJg7LpJQRMRYRuPWxG3zLy76GGOEtb7wTP9MG+JIzJCQX1WAgCoPtKa98+WsRga/62uezvFGghIaYNojJjqSlBGIMTKcTdrZ3uHztGrMq0OmtcuLEWdbXTpDnjtFwj8HBNvPqACsL2q0Z0V+mnO3RKzrsXzXc9Rf7vPftF5lPPGdPtdlYdfSWob0C2ZKl6OcsrfVotwO721copz5Z0Ko4sRSFY1FWicMxUDccncssNk9PaUmBDY1gxeBrz2JeYaVDnmVk1qI2Q4wn6hzEp03JpkDN1UuXsQLWCcwj9bxkaalgZX2ZPAdfLQjzOZmzZK2C4ByL4JlUC1Y211ha3mA0HjK4tMX6WptWbpAIraJLXVoGg4pW0WVlbY3xBD7y4Slv/fN9/vRtu9y/VVJVTV6jCHJ68jgAACAASURBVEYyQAmHOY5YrIFTJ3I+/3M2ec5zNnj6Z62zKK8SdcZoPmN7a4f5cIED6nLOcitjY9XR7y5QWYCNhOSAkudZivprbCL0QtHKEUkpWNYJWocmEu+JKgSf7H2XOBRs5hIoxUiWWyKRqBFL1oBV8lKsNRgRQkwpZZmxFEUrpR/VHu9JwOZyIgreUxtwzqSUpghlFfA+JKvRWgiJ303OSIroB/WIBWct4dHA6VW1EqWJSOISUWwtxjm09PjgWdtYp9dehmjQoJTzimvb+4wmwmAQ2NmdEcQkZiB4Tmz2mK6M6HWV0yeXaHVXECMUvR5Ft8PBcAfjLHkrw7nG7IeUZiFKVQNagOaJJPepb0YDzsBkMmJwMERMztLyOidPnaMqK6oypCTgKERNCi8P5MUdppYYgigmizz+yef4tu/+OkJU3vqm9+DnkaCNVZic0ibQkYBvf2vIb/7X12CM4YUvej7dFYNVRaIhBkHEMZlO2bm2xeVLF7h69QoiltUTG5y+/TZOnLoVQ854csBkus+iHFL7MSILcjthPt5F6prpMOO+e4Z8+P37XLt/RjWrWe716PYy1FTg2ti8xXwx5+q1CQbLyRObbNw647737yPBYE2WWDcxOBuJIeV0iRSIBKyzSFCCBGJM1oW45EqFEPC1Zz5bIJ0Max2qYKwhs6BS4rJkeVF5qknFhQtXMLmgtSfPCjJryaIy2R8wmYzxAtIuiMZy/u67Ez0gwmOe8Dh8WDAczui2l+i1W0xGe9RlRStXcrdMbAu+NmxdDnzgrgWvf9MW73rvAYNp4p+lyStFUloIDUtlsOQ53Hpbzt/7e4/nC565xup6oNsPuLqL0ibWgW6rJI+WjBrX8xTOk2UeNRBIY2WQZLk1oKAacc4QYsTaRCGHGDEx5Tke0h8hJHBxJkVuMysQNFlzNgW2UEUQvPdpLoylKHIWZZn4QAFjHc44NCh1iE3iu6VapEivAHkhZIUjkPJYrXWJx42J29UmN7YoCpCI9x5jteGjk7fyse94vTly00Gv9hWYkrIuCT5gRRBNKRg+eDAGHyFITowZg4Mhw4Mpe7s1k4lhcBDZ2w1MZlMAOq0CCZG6DMx6AR+mrK4J/eUeuQOvilhHiCm6Fr0md0HBak3LFdTBEeuC3PYS36aOLOtjTReNMBxMCTFjZXmNjY3TlGVkNJphjUUPIwwN3yZiUjTTWrQJQIhAFIPB87gnneE7/vHXowpv/5P3UM8jRNPYoPFjgC8EYW9rxKt+9X/ibMZXfvXfotM2LErPtWsH3HP+fu6++0Nsbd1HVc/YPLHBMz7rs7nttsehRYvtnT2qMgWIqjim9mOqakI9HRGmB7TqyMHlMR/54EUuXTygnEU6WZellmClpK4i02lFcSDkRhhnFYYZs3HFbFrT7bbpr7cYXw2opiTgRJoDrkU0lkyhrubEOmKcS7ldkmFIuYVoIIaSspxTV0IIQrvdIcscGgwqDmsFa5RWkZPhUR/YvO1WNs6ssbN1kcGVS5STOdujKWX0RCecuOU066dO0F/pMp4MOX/veRTl8oXzdLst1peW6XYsJiwIi0imOS3tMhlWXNsr2d4teP9dY976jn0uXFowPwxWKMkahxQZBywOJ7C0BE9+2hJf9PyzPP0zeyz3Jogo89LS7vbAQwglZ0+dYLh3mWo2pXCKMxF1kaqJwmvQFPUWIUTfBC0iRcsQAik8ZSCEOnGmRpDc4MtAWaeEfJvbByL7BiEaTTxzo6sxaoqyH+ZW2gxjUuDD2nTiJwZJBkIQ6jogODJXsJgmF9V2LCHMCCGSZS7pfkz8n3HJ5VVNAOhDDQQkKEWRDI8H6JqbLDcd9MpygZiSslxQ1xFrkvVSl1Uyn6Nld7HHdGbBtzjYHjPYnzHYqxjsB2YzpfaKDTkghAr2rk5YjITlNUtdRsoSFh5MXjJf7OKsR8QTAuRFRpC0YzqX0iw0GmKd0ess0Wot41XQWLCYRcpyjDU5rU6Pzc3TWFews7MLCM65JmirKSJm0kmP2AQWpFG6dEvKrzIm8MSn3cI/+p6vJ8TAn7/5fYRFSilJNl9scvES8Hlv2L5/zK//0v+gGi+49dZN7v3IeT5y7yUOBiOMjeQtxeXK/s4VLlxwuJahokuFxWSWIHOG421Gw13CvGSys8PexSvIUDjYGlAtPKgjVjW0S06udWh3LJPFhGoWGEmJxkCeRwwRX5ZYhpzaNJw+vYn1E0ZXIz4WGGsIRtJmY0A0YKyiWhG9RYMjy7pYk1EuFkDEZTnKHIjUviTzebLIxRDDYV5iJNYe75VgYDoY0u1lhKrGVzUrvS4rqyuMqjm74wE7e/uMZmO6nRbOwVKeM51NmO7t4WKfaqZMbEwkfBXIMHi7YDB23HM+4y1/PuDdH5ywPw6EeGiFS2OTJ28lSiL+u7nh7JmcL3jOJs969gYnNmv84hL7s0Cn3+NgMqLTWUW9Us1GZK7GMKHTA9EIIWIeyLlMFpx6xWYGYwxKaI6WpRStQzfXuByTC0E9RCVrp2R+v1AWtccWGSa3qGkMCoGgissyHBZfl9R1ypGtq0WiC+SQlzOIWjqtNlEjk8mE4BPQd7utdIInLjDO4FxBr9cDhNlshjGaPJzgiSEyq3wDqJDllrwoCBoIoX4gaf5myk0HvZ2dfTrdjCxrU4Y5oYpkNu0QMXi8rwm+ZvfgErOBZf9axd7OgtGwxkmHzPZTwm463kCmgvXC5GDCYH/K3pJjY7PNeJDTXY20uh61nswJahw+GOoQQQxF3mI69UynByz1lpGsy3QRCEEQPJiI0RwpDJ1uH2NzdveHBBXyLKVjRGJyaK0BMYi16eyuNSjJ9dCoifNDwQSwFZ/xtHN85z/9RjT+Gu/80/c/AHyJ1k3Z/BCTRRQMO5cHvPy/vJr11Q7dToZzls5yl3annazbpRa33bHC6mab0XibKi5Ta4fBZMzu3lW2r1xhsH2AloFqMiWOPMwXWBPIKIh1gOBxrcBGd4nNzTV2DhxbozEHB3MCNWtLLTLbYjKq2DYluZ1ycr1gfaNLPR0xO6hRW6TjV6bASEb0NcmSsGSZadykjCgWm7cf4DIzl6FapoUZ5kymU/LckuUW9QFrDIP5hEW5QCwMh9vsbd1H0RKINa6dY9s59WKMEeh0CzqdNr4smU3mKdVDIzYjpWMYg3gYDytCHTA4FmXB++7yvOHPtrnnSsnUp8hzk0zTzExMZ6LFkjnh1KbjM5+6xNOfscLnPfMUK8uB4d4OJtRYDPPRkDifU1YCEWbTfTJXs7yaURQZoY7EBkqdWAKhibpG8Olkkdh08kI1IHiMTSc8TJ41p3Qi1gXa1lIUlmqu1NPIoq4ISMosEAFjQYUYDaH2gMXalAHgGzpCg6MKBo2Kc+mUlEjE2gyIiMmpTQJ/FzJUI9baNNcRojcpt9WkbUI1Hbk0xmGNSalqVUCcNJGiR4F7m2VnqauKzBX02hErgaqc4AOIDVBV+GlJuRcZ7xj2ryjDAWR5rzkalsjS3OQIQl2WuMKR2wIqYbw9QxYTsmCwPpCroegbcltgJMeZDp12l6iWiKWqSpzLWF4+h9JqkpETeSwieAVrLK5osT8cUFY17XY7HRrXQGYNoSHl1STyOSlicndVU14eknLsEvmtiAk88Sln+dbv+lqqqubdb/8QoSJlqadj6YksxjdegDIbLQhlxcbmMqfOrIHUDEdDVjZ6rG2ucvLErWycOsmV7SHz6YSdnV0uXdzm0n0X2buySzmtUQkYo7RbhiIPtKxJxHJMnFa/Z+m1cyQIC19QhppZVRKG4ETpOEtlc0Qiy8uOzE0J8wqvsIgzHBnRKjEWZLKCmBy0jRpDHWtcy+KrSDCHL4CwZNZijRJjSYwLar/AGIhasqhSBBEJuEzo5TZtMlFZyhxLSx3yfou90QHb4x1avZxzt59M7l/lWYymENLCj6Emb+esLi1TLuZMRlOit0Rd5dJWi7f9+ZB3fWDI9shTNi8GMNpQFpoOkiVnzNBrCU96Yp9nP3udz3xGl9NncjrditqPWd/sEaeB/e3dBMgxMtzZZ2VlmdvPnWY62UPUo0GbfLnk6hmRJgLa8LwmbX9GQUNK5DWmOVFjQaQmAVdGlmdN2ojQygwLUzEfgw+BOgYUwWUZGiCEQPAx8aokLDTRpNSVMiWCq6bNej6rcZlQlZG8vcLmLU8iSkhueK3sXPoAZTmhyGGxqKgr8P7QNQ9YJw9YeSLp2Gm9CJgsRYfrR8PZ27Pnnomv53Q7Disl5XQfbU/xi338YocyLFhkQtHpIS3HyEFddMC0QT2192SSs9TpIAKjmcN7j7MtcuvIiwVtN2WlJWx2PJ1sTjuPFJ02tn2CrLWG2C55Zw2kzXha41ybXm+ZcpFSXuq6xquiJnFKLmtRVoFIJC8yrLPpJAGSuLpDS1UNQSWdLyTxGYmsTa6vNm+0QCSdmnDKUz/rMbzse76eX/i5X+N97/wIsbTNGUlPo/KghynNlqpU9vammNzR7gfyrlJ6z90f2eL8PddYWV2j8rColZ3tEbPhgtHemHpYEU2gu+ZY2cxQVxKDEqpAhqW9lNFRpb/apQTuvzbgvmsT5pVHjWVaCcNpTVsXFM7QWzIs/IKNNUOmngIl73lOnc6R3DAYTJgOK4h9lLTLq6ZotnXNueBG37VJ37FGsMYSYkBsJDeWvOVQUxHigpYTTIRYRTqdJZZWljBZzXxRUvmK1ZN9Nk5u0O+usH15h0vnr9JyOavrS3T7bapqRogVi0nJaDJlPIfRbJkPfxj+7O1bXLhSMvX6wAsfUmaQEjQkblaFzAon1zOe9cwTPOvZa9z2GGHzZIvdg6vUBPpLfQZ7c8K4olzM6HUtzihzE3A2MpmMqcqStouUZZ2IfNUUVdVIjCkFhobzqquAtYKmmBiQTtaEqERTJQvKtlOeZPSYhl6J2rxEAkmnZiKEukaa42iZzYGU3JyZdDIjxCblidic3FBiCPSWupAVzGrLlft3CCE2RyAVH3IyV+B9SAnntRJDRCRRPjazGJvO2qoGrM3QaAl1ykFUeRSAnrUnKQrh5EafVh64/8IHmI+ukMWMuFBaEVweiHmgrCLTecQWG1izjK9nGPW0c0unlY61BN9ltvApkThvo2qZlfvMF1Dk0M89NhpMqeS5I9Q5rWKDc2efQhVydGdMXUem0xpr2hRtg8tr5nWZYnLGYVyeombO4myzOxrbJAFbMBaxkvgXaRKUaRKUsQ8cPheTEky0SW4QI2Qt4Rmf+3i+/btfzM/+1C9z7we30WAQXHLjD3P+iIlcVpjNlKtXDtg802KtaDM9mKe+qrB3bYoPwmxaUVcRDVBNUwqJWk9vucWJW3osUCZTz2wYMLFmqeNoZZai32J/MuXKYMxw6pnO0llhsYAEDJG2dbQmsDee0esoJ9eUU2vKbacLuks1rXakv2QZ7UUOrg2pF22qyhO8p9taSm+ekYgxBc5AniWTpip9kzbhktVHpI4lrRwKAeoafDr9MltM2L88YTwvEScsbRZYsUxHQy5++H7GOxVhAUsnW5TzktFkhK9rnDNUUZnVBVf3ct7z3jnvft+QvbFSKQ+kQqV5ShwYgKilsPDYW1t88d85x3Ofu8nGSY/YklahdNuBvN3F0aeTK52TsKtjiHOszTh1coXpYsFwPKXdBZX0Mgn4qL4Ym2L/IaY8Tg1pc3AuI0Sf3rKDNjyawXtPsIFKS5x1qAg+RmofqH1o+m5TKlVMZ129V2pfk1slKzKiB4ygXom1Nm87SulUxiYW8+BgRN5tE8KEq5d3mE0DCHR7lhMnOwQiB+MhmXVNWhWNvguhSsk36bhlOpJYLyrKEDCZ/H/kvemTpXlW3/f5bc9yn7vkXltXL9OzMUiyB8YaPIDBIwHCEEbgkRAQoQBFyMYR+gscDkf7td/JlhGCkBCyrbAByxEyQpJDFgMMzIAGhplhpmfpraqrqqsyK5e7Pdtv8Yvzu9k91vsmYrgR1ZXZkZV587nPPb9zvue7XJP6383Hu/4ThSwiiXYhjKIlzcRb6ZDk5CcJ4VPLnCPLgczsJ2MqCiNYhZLBI0U5obTSGT+QTaqotfT1ciGlQAgD5OeiyPKta285wQzFmeI6UT13d/LYLTDe+W/k93v7v28Hzr/zkf69T72PXF4s6dp3hMSl9I6v/ca/lQJj5XdM+XrsliZKi3WQdRZtdf69d1xmuaFj2Il+5enrXDBTAKJ0WybjP0oJ7KKNEKFFbpawJlE4KEuFtcLvEm2v0HVCgHFMxOwjpJRFa5MpCur6j9gnpmwxJV2xXLcMqmeC9/VVTjtCrcZaS1lYwZ6SjH9Kaaq6oqgKjNXZQKLAWpu3mQmjxfWmLCKzmaGuDGa3TOQbgfV3fhYTtF1guRzYbANDH+V6xoTP42ci4oPPLjbk7yjaU5W9mL7BrVy9/Ve2Obz+H0rv7qOda87b/273PdQ7/+3u++6WIrt7ZvdF+d5RWuXt6f/v/lI79czu6+V3MkZfvy7GWJwrcK5Aa5sVNjLtxOza846beGfyA6QsoQS0ural+tN4vOudXtu9wThu6VuNVZ4wrohqQ59agg10Ctpe03YJbQxVobhcLjHWgw5EHVh1oHQDKNbtFlREYRjHjhha5lONdYpN50mdpi6htAGfthRqzTAm3rjXEinYtAFra/YWhwx9oN140YoKUYyoLdGXWDshpsg4egF/d/582dEkRRF0GyvcppgxGimqZHKp9BEmE5ZTgO3a8zv/9nP8o7//Kzy6dwZBo5PoH8lU3uu+IxNSJ03BzWf2KSYjygbmh3uUhaOqSg4Pj9n0nrZPPH7riqunG0K6xF9siNGyuvA8eHWJsgMasMFisfTbgMGzHTcczvfQhzX4DZthwOMxBRTWMCFSOMV0rrlzp+DoQGP8iPHQXo5cOgMOlsuRflOj/YzkZSzEDPJm04aYLOmaMZGwRmOsJjGQGPObQqFSSdcOhJioCoexjjQmCltxfLCgqBOjiTxZPmX0nkW94MWbdznbO+PRG49YD0vmdcP+3pzRFyjtqScFq/WKWbPm5HDKc3du8pk/uOTLr2y4bBVD7s3l/ZnhBRUZo+Le45F//q/f5P6jLd/5nYd84FsKbjmFDzXbdqSerogqcHbW025GmknBMCZOn15xcLjP0f4J7fZSdNJmyAUrkpLOhz1ymMlejKQjqCBuNSqhrSgrQkzoQqGNpXCVaJ4z/1FbQ7KJqAM+MwmSlkPFGoXGopXIFEWqK52gTorkZZGikBHbWMN0NmHTDxTlgve8/2bmDkrx3m4eodOa/f05V5crmWbsjocAymmUAWUzNqkjtjLoJJzEPxOUlYcPPoO1CT+bEPxACiPBd8SxI/qOftT0KbIaO5Ze0RPpwwpSSV3WFK6ii55Vu0Yu7EjlLDGt6XxLjD0ViuWo0T6J+eOoKbY9ZnyK7rcYW4F2WFewWm/xo4L0Popin34cUUnkaRqhpISxozAa7Uratif6iLU7QF0IpEZrYchr8Q4TMHrXKcbcyQgWQowoZfG94vd+6w/5pX/wa9z72lsor8RQNRc8QQBzh6pE8lTXlqPDhsWsopxMGEKL8pG779nnA9/yPPtHxzw8W7Hcak6ei7z55hPuv/EG6c0nbK9axiGyfBJoyhKrB0ojXV7XRqKOrGg5aeY8f7jAac3D5Ypl21E3ihsHNZU1GD0ynQXu3i04nDti59g8DSzXjvPHVp59arB6jkoGrQZU6sVGyiviaLLjM3KNlKgJQujyMsPLuBtHhn6UDs5oum2AFLFas26XbEPLZFbgJgU6abaXA/c3j+hOPHVd8Pz7nuHq6ZKh61ite7p+QzWxTOYFk2mF1j3OLPnQ+yfcOTnk838y49Offcrrjzu22XopJVHQqiQmAjEmzlfwu//unPtvbfjoXzzkw9825bkXZhzvTfDjirr0TPYKTIr0w5YxRGxZE5MBsjciUFUNMQykIFOKIi9NlBRColA8dF5yxixj0XKjyQHATuoHxogRBhl2CT6IGYQGlCwyghdblTCKoWRKhhACY5ZQ2tJkY9iEc5bJpJLuMAyUZsNsBkHnLnyM+M2aFEe0rmiaCdtth/dBdMFJEQkyWeROX4pwEDqO0RjzZ2B7a8wlVW0Z/IZ226NQFEYT0oCPiWgKdGNobIWZWHTdY0+Fpzf6gYhDOc2Y7YtcofExEOKAqT0H+xWHxxMOTiqquUdXPVHLYoI0opRhO3SklKhSQVUp2hRYXj7i8LCkdAXjGLO8CJE7pcTQtdw6PKS0HW3byk1UWLGdykORJqFiyAD422Ol0FWEfyejpCH0hs/8zhf4Rz//a7z28gPSKCehyQTlnb4jYEho0JGjk31OjhuxCe86sDXzvQN83PLowVMiWw5vHtDs36WczDiaNJQHc46e3+P08UMunpxDH2gvrmifbGGTUH5EK0OiZBwC621k2Q6UUw+qw+qepob5QlHWEatHSjswnyumE8/+fA9vNZsnVxSqxKZCfn/t0GhiHIh+Q9IDWhn6PmGUQSWdNdURP4hmNDGilSIlh9ZQVTXOCRyhNPTdQNd39Fr8FY0tZDS+uOLZZ25zcHTIo/NTHtx7gnWKelKigvDBDKCswqfIxXIlXVIBaQio4ZLD/Zbv+dgBd+/c5t986glf/PqWq04KX8yvsegkEoFIO8Irb/ScnT/i669M+fB/uM/Hvqvm5s0ZlxcPGDYbnLboqmHYrGkOFgSfePDkDGdH9ksn3ykFYpQYAYFPEiFkA1Lz9rCtlBQ1uae0qCyiAQM+DvR9IEZDDJp+k/BtwqIpCumg4w62MQlTWLRRjL0Yggr7QAjRxqj8XMSBfNIUxBQoRo0PW87efBmfy4ZJI0Z7bGFwhcbagpAGbDCgPCkqQswLkQioiDGJohSMOEaP+bOwyDg8PAS1Zdt2FKXGKCXwbDTEpHFaHHGPju+QxoLp7Ir5QcvV+cjluWe58nTbkSHF3Ehp6onlYLZg78CyOLAs9msOTuaYMrBcvSmdhvIYC9YktBXNnzaSgzGp5yRvUbGjKktU0iRl0LbMm9rIdrOi7zbs7+3hx44YRojvQIDiDveTzaxW9hq/2HFWRJ5riaPj9z/1Rf7B//RP+fqX7sOwK3ghYzfy9TEbUCqXeOF9t/lb/+UnuHvnkNdffYMvvfwa9958wOV6jSsSqQ1w6tg/WXDj5HnaUNDGxL6bk+wBq+0+q+VTfLtlfXrK8vET3Frz6NVTlpdbYtCk0RKt43Tluewf08eOZBIHexWzucW5gFYjewvFjZs1hwczRh956+El/dZSGoNVEacSQwrZ8y0wxo4YeoYommfjZDEThz6PN5G+b0GB0TXTSU1ZlRiTQI3ZMSQwbWaEcU6vI4fPnXDj7glnT97k7I3XuTw9Z9x2dHFEa01RVlSTmqapSMnTbtes1kuGEJlNZxgndvL1zOK7HkeBM4p66pksDrn9+Smf/qMLHp2ODAG8sM7QaGxShBQJJC6vDH/8R1vuv95x796W7/n4bd733htU8wtx/JnU+NIxnR7gu0A5KSldwA/n9F0noLqIchgzZCJO2Uos/ZWGGBhDAK1RCHPAWs3YDdhScMoUFX0b2Sw9DIrSaqrCUVmxhh+Tl614ihDFeWXoQ8YeZQmndnitkyeUkmf0bTac7UWvHm02lUA0wnrEJs8wWmnaVcAV0r37UXS9WidiJlAbdNbc5ufy7te8d7/olWVJUmI+GcaA0QpionQ1bTvivWIxO6SuKsLosEwoi4jSI66uKC7h7HRDbDsAqtJwdNywt6+ZNCNHRyUHhw11U+CqRF0ds1w+RetAPXHyJtNyAmkr7HCVKkxR4nSg78/x3mKKOUXpmO3N2W47Nu2ax48f4Kzm+Gifi/MLUghoazOKHIUwreSNoWKUpUnkmqSskiaOhn/36S/x83/3f+OrX7gH479f8ISua8SLzSWe/8AN/vbf+XG++z/9CxQu8cEPv5ePLT0PHj7h5Zdf5pVXXma9ueDgcJ8bt97D/uEJNxYzxqDYblv6fkVTJ473K8ZxQ3hmQdweUvQ97Ydv8dUv3ePrX32LzTKSrOV0NWCtYv9oiqtaZvslJ8c1zrZoBg72Hbdu71OYgle//JjuSmFlly07qjCgUyRqQ0gKP/bYIqFNFLcbHzDKoS2k5Bl6GYmcbSiLCmPtNQBujEKbhCs0VWUYuoFuDAQS3TjQDwPKiMJmvVrS+YGoItODBftHx8wPpqxWVzx8eoqtKo6ODsEkXOOYTYU29fTsITF0uLLhoHb4YkuzX3H3uWf4nd97yp98bcW6FY+YhOhuDXnBpgK9h8eniU9+8ozX3ljy8b/8DP/Jd9/ixnFEm0DtNK6qMBhCt8e4WXFx1tJtOiqrcVaBzty3vDjQRpFiJAayFMyQguhly9LhfURpn62ZFIwK30YYE85oSmsgBtmeaun2ZAOsSEGWLSkmtJFljzGOYRhkKTNGOXK1ous3jONIjAmtxVxvtelQGhZ7lqQGQhjZbK5wNqukvCx0YpTNdFUWjH4gIQqNoR9yR6n+VIqeeemll97VH/j61//xS9LOOxkdlUYnkRvFkBgHjzOO2hVYFakcNJURk8LKYFxEWc/eUc3+UUnTRE5OCo6OCw73HbdOpjQTg7MJhadtN7TdGpu3nc64HKxjcLZGKYezFWVRgOqIqRWqRAgYV1BNGqbTKX23kfE6BubzKWXpGIY+bwTzaXtt4imjhNgkZQF3jMTB8Ief+Sq/8D/+U17+/GukUQwYdyMt+V8FNFEpcPDse2/yMz/7Cb73+74dVyeijiijcE3FjWdu8uIHX+SDH3o/zz5/B1daVtsVV+srtNUcHB6zv3fEpJpilUNHjUqaqiw5OqioJoHFgeb2cxU3n2nQ1rFcdlxe9hhbMp87mpnm4LDk8Lhgfx9OjgwvPHfErRvHPHjjlMuHLdoLtmkBq2EYOgbf+TnehgAAIABJREFUE+PA6EfQI7YQ7Cn4kRQDKfm8rAlZnlQxmcyYNjPR3WbpFXokph6UZ/Ad7dAxxoFNu+bs9C02l0/R0TObTzm+eUxZKVTyrFZXnJ6f8eTinK0fKacNQWmeefYu9XTC2cUTtt2l5ICkQYpNYbFlSVGVHB1VvPe9hzx7dw9jFFfLnq4HCfLLDoEZbJOBNzJ6xfmF594bV/hQMF3McJWhaSq22xVtv6brtrx5/yHbdYdKmnEAFTVWKQqrIHlsxsMSKRPyDSEo+t4TfBK3cORvY7RYOQ2JFGQTb42QW3aFC/12akrKaialTVZSCLlZil3I23WhVxmrsWa3udeQRINrrME6BWrA2ogrwDlJXksq21FFRN4ZEsZanNUS1ZB9/sjbeo1h/wP/7X//btagd73ovfn6L7wUvSeGgRg74hAZB8849AQvAP52vaHdbCGEa8C1aaY0s5KyjkxmmumiYrFXcHxccHJccnI0YzGdQAwM/YZh2HJ+ccpqfYG1cgOk7OorchrRAxptqUqDLQLoDlNoEo6kSky1hy1qFosFxhq6ThKoxnGgaSYC1oaQqRxJHCyUnJIqLzKEUpyIveazn/kq//DnfpUvfe5V0qgwSWXETlw60vVIq0g28dz7bvLT/9Un+N7v+whlI4aPCYhagTUEBIeZNhW3b5/w/HN3uXlyjLGG5WrD07MLxiEwbRYcHdygrmZoDCGAHwuKYoK2A1GdsdgPnNycU7o5j9/acnXZYnWiqTWTylO6jvkscXJYcrRfs7lacu/lC1yq0FGwMZ0iIXjGEIgqYZwhxhGl5JrHGIg+5lS3RMy25TFCWdZMJlMB41MARlAjIXakOEIaicCoE9W84b0ffJHjk3269RVDuxXQXGtsVVLU4m84Bk89nfCBb/0Qx3fusFquOHvrEXVtqSearl9iXWIyKamqirKo2LYtQzeCgum8ZHGgODp2HBxM6HvPauXxIVOrrr30xBE7EYkJ1tvE66+veHh/TYqK/b0p/XbJOKwwJtK2GzbrlrbzLNeKYTCUrqQqHDp7DqYEyhjB45JYM/W9J0axSSPF60ITo7igjIMULWukUzbZSU1bcQGKIZCyCQCIk4pSOn/vEZTw8lISPRA65vdIAoTsrK2hbEqKWqP1iDYjxohJQSQJd9DHTDjfWYgFrJXnKrLMuNtfoUgcfMt/981d9L728v/wUgo91oiOMEZxbA0xMgaxxiZF+rbHj5HbN55h3iwo6kZsqKyA2kVRUVcF0wksFiXz6ZR5s6CpKqqq5MnZY7btEluCcRCTXHhrNdp6XJHyyToSUycSOKNB1Si9wNQ3sMUMrQxFUbJYzGm3W4ZhYBzF+WIxX7DjGwp6n00ylTgwQ+ZQecUX//B1/uHf/z/54h++Qhx3tuIi4JeHWKMHpcAm7r7nhJ/+2U/w8R/4jygb8k2YSXNK5ZFZS2eZZDS2xl7TVo6OTiirmm3bcnF5Ttu31JMZ+/v7zGYHpDRl6Epi1CgVWG3OUNoz39ujrO8QaViuV6TYUZpAbT2NizjtiePA6ZsrumXEJQdDQidQKTKMwzUvUmslucDOkjLbP/hMUt2pjFOinlTUVYVz4k3nfYePLSHIJldiO0UvnYzixrN3uHH3Fl55nl6eYrQhhpLBG4p6StnUYDWbdkNKgdmswRnD5mrJ6uKceqLZP2w4vrmPsQo/jnmD3RHCwHQ6obBijV7UlunCcOt2xd6+JfjE6iIwdrvOSbznDGZnKkYC2k7x6FHL1792zlsPVqQAk8qxNy+YzyzGJopZyXT/gE0LZ+dbrCtFE6xSJoMLvy+RN7eZZ5liQBuFc5nAHjUxaMZB+K3GyFSjMs0lJXBWIkOH3mO0wyibuZ3Sdfmwc0WRomqNyMeESxvQJi9SrBEajQ4YnXBWCqIPgRByIp/SKPS1iagxCldoGWlR6LTT2geKwrH3zd7pvf7Vv/eSNcLs30m30KCMIuJBeRk5zZTK7bE3P2Y+XVCVEiY8hoBzVf7/c8pCMIP9xT7zekphS7xPbNoOTEK7gLIeYzVFWWKcXHyrNdYiUZEYEhVKTVF6QTl9HswMlRNLU0pMpxNcYem6lpiS5Ldqk8PAd0Uvk52VBLCoZCAkvvRHr/CLf++f8fnPvkoYQH9DwUuQPwtAsok77znmZ/7rv85f/sGPUs2UuDxntulOoqSU0GpiJhfvxhetxfVWG0Mzm7F3sM9kMiH4keXVmnYzolXJ3mKfvb1DqnKB1RLQ5P2IKeY889wLvPj+F5kfHDKMA2PfUmpY1CW10fSrju5qILSJ1O/sliLakC33YzY6FTBeYgOBaEnRoHHEAD4kyqqgaWoxedWC8YUkgdXOCStaZbJ0YSX/oR06NkPLuAtNSonkA7Nmj8X+EUVdYStL1IG9wwVDv+Hhg1fpVpdUTuPHgcCAjyNXV1esN1suzjfEAE1TZdNTmw/IyLbfoG3k1u0p+3sSX3B5uaXrEiHu4h7FNkDe1NIFhpRYbxNvvrnlwf0NmytPGDyl00wqRzOvqKYVtqx4fLbm/LJn9AbjnFB6TMhFB5SSw0OoKYqyLCR/InhiUGzXkXYTMaoghYhSYuSxu2e0loIdAhhtiUHs5wEJV1fgrFxjrRQ+iuuyNfLzUIaoHQrwcQSi4PHKCOFbCa8yRoghYbXLTYZCGURiqJQsbbLmWMjOhv0PfpMXvde+8nMvxdgKQ16VxCjhzjEFohrQukSrBYU7oSlvotIEpYQfh9K4qqRp5tTVlLqqmDYVhStR3tBuPDEWbFtIugJbkDSYIqBswFqHtQ5nSsrCCV8wjCgqjD7AmCOUvcFifhelpAtKSZFCQKnEZFLQ9y1+lGyP0QeqnffbtWIkB5ljUV7xlS/c5xf/7v/F5//gPnHYicpSpm7uCl5G9Wzi1nNH/PTP/jW+/4c+ljs8sQSKyM0bk/h+iAW+EH2T4h0dIDKmaE3S8ndZVcyn+0ybGSkG1ptWujg1stibc7h/h8X8FuV0jisanGtoZgtO7tzg8OYRxhjWqw1jP1I7A30ktQE1RKHoqIC1QqYVkbzwE21eSIQov6fWjhAgBEUIkemsoWkqtJa0OWOElC2fa4wu8uLDizpHQV0VJMSl45ln73J884TJfMLl1TnLq6ecn5/y1ukTrlYX3Hn+Ns+/5xmKWrG6OOXmyR4nN/aZ7k1ph57z8wtSgpPjE/aafYwumE4XbNuex0/O8zJJgQn0vmXaTHjuuUOee37B3kHJ1dWG1WrEe32tOtA5lF0j3MqkEmOAy6vIo4c9bz1oGTooixqVDE3TMJtPWCzmFK5mueppu0EKh5Eit+PmxZSwTsthkBdHCU0YoO8SXSu8PFHThDz+GumwQ8ruybviFPMYAkSFs5mfNwT5nhGSEk6hdYaEoe09oxeX6BAS3otjdmGlawxR4UdFDBqNgRgpCos1ChXy800xd4TZjENpDj70TV703nj1F1+SF6zEminWNCgKoASmxDhHqRMm1V2sOcQPjq5PGFMwn++xWOxRVwmnegoXaCYV0+YIzYTNJrFaweBLtFtgigafpFtwdkJhG6qioSqnWFtIAVEabSpSmhDYJzKjLPcpi7kUvGwLFcIgvC8lnmEg8GAIgem0EXA7B26nqMBrvvL5e/zSz/0LPveZ+4QRUD4TcTO2oVR2b4ZoAjfvHvA3//aP8gM//J0UjcQ4xqziSAnRSGb5XUwiJJfA7MwAzGMv+WRXSiqRUSUKeQM004bptMEVls22ZbVqiVHjyhnT2U1ms1sopiRVo6ylOZhw8/lbHJ/cQFcOpR21rrBjwkaPRg6ESMrYknR62uhrN4+U5Wm7N57Whul0QtNUmZYidCKU8C0VihQLEpZh9PgY5PAyCWMSMY50qw3L00v6fqCsK1BihX50coCrLNrBYn8ORnNxfsZ2dUVRatzEUM9qZtMZm6stm+WWqjAcHh3QDQEfFUo7lqsNk2afqlrgygZbaBI9/WZD9Cuee37Oe148QunEtg1sNhEfTbadirnwGRk/VSSqRDcoLpeRs7OOzcqzuuxZX7XM6hmzSYFVI1XlGP2S7aaTAjJK/AApSLE3Agns7Jp0MpAcRIMfhSZkcxiTj1lUljFBgbRlgSDduGL0ARUTKiS61tN3isFL49AN0m0XWhZUKkn4uooGi4Q7ORUpjIIAbae5ukqMfQHREkaPCh6LGJj67AMYYy7iQZ7L4be+u0XvXaeseD9FMyXhGAeDSg6VAkaBswVKlRAbYmyIqSRGT0gjF1dbBt+y6B1D/5Q0PAISyt2UreUqsl5KzCO6QEdLSIX43pkDrOmxqkfFRPKWMYwoM4U04gPENEOpGpsmdK3kBTjjwMiaPcVEtw3U9YyyWNN2LTGObNsVm03JYm9OuxV8UgXD17/8gH/y8/+Sz336PmFQoEXrG5FCFPGolHEhEzi+teAnf+aH+f4f+ihVEwmMmfMnvaEyYuwoQI1I0yQsfLf1j0RiVn0I015c4CxKGTAZoI5CiZhNp9RVTT9EttsNV08uccYxaxbsH9xikTzb9oLWnxPTmtliwe33PEu5uWJ776ssX/8ycWxRYchPIOfqxkRVGELSeZxK15mrISSUtsymNVVVkpIXagZKlBgpkJJGU2FMQVJgiyLL1QKjBhUDzaSkiZr1VcsbF6+zuDqkaGQjf3TjiBcWc04vnnLv3hsMr3mO9qbcfua9JDPw4Mkj6uWaeTPFKU1hC956fIUtJ1xcrOk6z8HBHtNpgx87hq7EjhO264g1kemkZLk5Jao173vxFnfvfpBv+7Ytv/Hrr/D5L16ybTVeJaLy6DzK65hIOhCTpe3h/qNAGHvadUG7jITtGbfv1FQ1aFdS1wdouwbjGUIibcCkApcCgYhPQlyOIeSRUeMHSRPUVhGsEilaSnTdkGlhShxOAOusFFKTYZYEYQiEqBlSwTjU9KPGpyV7e4Xkk6QBqxKDH0ijZBVbpSRECIO1JZDYtJ7CzEgpUNiADR0uRbQTpYlyGqsVVlu5F/WfiWCg9xGGDp9E/pSCyK1iMoRgUKoCle1nkpGQkczqHvoNy4st08YT+gFSxFQjm+1bhFiTKMS2SCOeYLpEOQOqJoQObQI+bwuVVllvHbFFRaSSJUaqGUfo2o7JxKCcw2MIcWQYEk1T0DRzhnEQLW0cWa4uKCtHYQuGree1rzzml/7n3+Czv3sP3ydQQ15uiORG4fN4a1EaTu7s8ZM//UP80I98J5OZygHRb3duCZULpFwzbWQu2YnMlYooLdperfQ1zkeyouYgQ4JZ0A9ck1FdEanqCeNY0G47LpfnXF0taZoJzXxKY2v80NF3K8Zqi6pqpn5Nv36LZXdJCj5n+Ga+mRK3kOQj4Ake/JgwVuMKQ1VVuEIRohCTUxDJXkwapRNaFRhjMU5cbIigQsxJX56283TtlrqwFJOKuW64fesON164w3J1ztPzJ6w2K8ESfWLoIklVjJS4oiSmcy7O1xw0BxwfTHGrJWeXl6zWgf3Dmwz9yNX5BaQOP3bUxSWL+RFFhPPTp/imoTRWMlrCmpPDmo98pOBg/0UOfv0NPvOZUy5XWiIydc5NTg4bZbsrvEXFg8cDm/UFm+Ue49bz1v2nDFFRzWqafcMzL9ymmARieEqiA2PQqiClER2j6MOHSBh6hm5kcxVIUbG37zI5XhOTmKUKlSQR4oiAfJlPavL1jYlyUjJqw/zwJse3P4Ch5OnZPfr2DJ8G6VgJGA26zPnNShLQSAWJCaYseP+HTtg/eIYxbXn64GXonlDWER96rNHXnepuAeTD8G6XoD8NGdoR0UWheATwBPwY8Ek4UCQZDNCQfMQaRe9FgmZ0RGmxhI9+ApkKEqMRuUscQQ+UrhK9qxb8S14YKyCqgsIZAc6NKCiM0RkfcWirUGpg9D0xVBgNQSWMsYxjpO890+mU1fqCvu9ziPLIernk5OAmr776Fv/45/4Fn/3UG/geKXg65AAgZBwESA6lNCd3Gv763/w4P/hXv4PJXJE0wv1PeXmhhYcVM/CbCJiIKD7I/EACWol8KQaNNcLCF97gzrXkGx9KCTk0IVxJWxYUhaJpSrptYL2+4nLdYUvLbLqgmd/GmoBvnzLqAGf32L51H6UttdPsknxjioze40PKC1fLbN5g8wZdSMeBorDX1AkUlK5E3EgC2auEhGL0HRBIcaTvO/wgtCPvBqroiU7z+NF97LSkOVig1lfcf+1lFJpuTBwc3+KFF99PPTMk7dms1jw8X/Hw/hnP3LnDzaNnCWPN5fmavb2GO8/usbe/x5N790Al6soQh6f03cCksIzDlqAsqSzoU6TSkflByZ9rLJPqLotp4vd//4KHj0c6L9kUKiZMShgCBqTwBcX5KvKFr17w5MmG6cTSjy19isyOFvzoT3yUm+85YAxvkPzrtMMFq3aDJVI6h6JgHAfazcDQBpxpJCgpjUQvBrfa7Lh2BnRi9D4vEjzaKnQUpoG2ooOtXc3B7ZvYiaFdX9CHDWeXK1TsOFhYysJlo4yYD1FD0o51F7m4ukTZCSd1Q9QXTJspVt3g6tEVWncSWE9ESfq56IV3rivv8uNdL3rKTQTgTRCjtNgqjag4onTIQKfkb6KU+PdH4X0pDUYXjFGj1RxI+FiRtBWWVBJSsfYeXeUcA3YxjDLmKWVRRiLsdF7pK6vRaZRtrx7l+cXEOIr+0zgJS/ZhpOt6FlVNVTmGMVNUEoxdy5f/+BX+11/4f/nMb73C2IFSPUn7vNcTYFsBRCGc7t8o+LGf/G5+4Ee+g2qm8UmcMnYGjDHKTZH0LjVj1xUFwXVipjdo0VAqHCkZYtRvGx4QkOK4K5Jyl6VsicSO9pKXL84m3KymrieshzWrzZpHp6e4yxU3jg6ZlXOagzssj54l1l9Db9YYYylcyTB4hm4gJnEBcYVjOp1SlhX9sAGf8zDyz3bOXY/cQz8y+hHvRzG9zMTfGH1Wu3gxqIwSBqVVIPQeg2Z5+pj1eomelBjV4vQghpg+slmds16eUVQN55dvcXXxlKao6LZbXvnqq0zqBUnX3Di5zWwyZRjWnF89BhKbpceqQqyoUCStWK0HcSzBUMxKXF3jw8i6PWe6aPn4X1pwsGf41G+f8caDwKpP13GONm/stZLv59GsR8V42mEVWGsomoYPPPcRPvzRn2L/RklSD1lefYmL068R1BN8e8HVxRVOQekmQIsxgxDvS/mufoyoEKlKC1HRbcZrgrFQJqUQ50wgYor42KOd4/LsNcbwKl0f6YbIYr+i30aWmw17psAoRT/6PCUYYoDVpmcym+HKgvXqCb4/o640VgPK0w+9vJ9BeKzCeIEkI++7/XjXi552uasKSMuOxqpE8p6UxMVC6kjE4xlCj/ej4FdeyJHa1JANASIS5O39SEQJ49+scVSCZ6FyqrvYYzvr0NqCiSSd7aGczqNDyCePWAiMPmC8x7mCuq6o6wVPzx8To2fS1Gy2EkgdR8W9157wa//k9/iD37qHb6VbTNpzbQKqpPQRHCjH3rHjR3/iY/yVH/0Yk7kItcGI2DvmxYQWjSJibSeaTK0JSZxpo/JvS3mSkT9oSZ/yZA5WtrBPO16f2FopLTZEkjUiI49RGimtCmUKJsWccjKhbTuuLpd8/ZVXsL7noI740VJPS1LnoE8M48i26/MSQ0lXgEURULpH64FkJX/E2RJrLX030HYd3o9470lE4XJp+f1Dpn4Qo9jom+ydpxTaGIwVmoWzkr06mRQ08wnajFwuz9HWo9SSN9/4AhcXpYQoxZ4hjExqgx88p0/ewifHZGHwPjEMa4bVhqvTlqePPdt1z2wO09mErhtpt4nJrMTpEr/19KueybwmkZjtOY4PC+YzxbT2/P4fLPn6Gz3nl55+UPh3hD+lTNj2ZAeeCCoqCpMo6wMWe7eYTGuSO6FavIe944csz9+gXz3hyZuv88XPfw4TNjx7a4/5JIDqwMgBJ0Ymii7K9TTGMI5Z/WKkWJmUMFoz+njtr2e0mEPoZLGqYDaZcHzjBtt2zemTkO2uRBOtlMLYEu8jzaLixs0TnC05feshmhYdeogRqy1eiYFpIuGccHIDinGQHOd3+/HuKzLe/ORLsulMmJ10K41oE3bqLeF1hYhKibHvid6jgtwUKmlIjvVmpOsixhREL9hQirk4mJwHkDlt1hqcUTirhE+kk+So5rg7bXV2cJVCQBLzUqWFViImiolJUzOOHYnAZFKxXW9IPnH2cM2v/vKn+Ozv3GdsFagxF7wsUc8nO7EA5ZgfWn7spz7GD/+172J2UBHUID8rv/67jdtuLNVaUrd2hpU7jlOKKZseaFJ6ZxylPN6mJqTs7ScP0VFmjE9nU8+UP1Za8CqTIwRRFNYxnUyYTiekGDk7fczV+euYsKFCEdst0XvQhqIsMM5cS43Ay/VIkk+rKAhe0W57NpuWtu0k5lAL68ZaIwuRkEgexjHiQ7ymaxircySk2C6hhWdmy5qirikqx8HhFFKPUj0H+xVNDUQvtKZ1i1YeZ0VOVTcNysGQ1mz7C9p2ZHnWUpo5z999kf29fXzoafs1GMXx8SGzacn5+ROGsWfWTFBJcfr4CU/PzihLTVV5jg4rTo4arE7URcnQjQyjYHpht7FXyPPPH0cldJDLqwu00Tz7/HPMD47QxZRqcsL84FmagxeYHbyPZvECVyvHo9MNo9Y005lYNzEKzSkq4gASgmauLeZD0kjyo0JrSwoGMBkGydw8qzHGSWJdP0rQdwis1j0X5wMxGvpe8fRyIGpFVVaoqBi6FqsD1nisCqToswJE3F+IQn3p+kSbOY7Rw61v+yZXZLz54JMvqShdhUxVQuiNIUjXFGW8TTHhRy+pTQF0Mpig0MmwXA48Pbtis+pIwVBahwqQlGTNCk/N5IQyIUYqElZzzVY3Jjv+agHed6HDSml8VoYIWT3IkgBP263xYy8FsC6JY+TBqxf8s//l0/zBb99j3GpACp4sLTRKxeykUaBwzA8cP/IT38F/8VPfS7NfyumsQpatSfHZ/a0zsTTl7IQYY45E1PmElIIl7hg6U0PIN/DuY7nu8drT72134m9wrt0duEbLNUwJgxxMVhvJHzGaaVOzv5jT1DV0oMfI2LWk5KnrWq49icI6oUxrL5tqBHvtO89m09G2HT6IYL6sHNaqnOUacrHX+ChUD1cWNE1JVTnRX5uEsxpnDZNJTT8Gnl4sWW23dMPIanlF3y1xOlJXGmci0Sfa9ch206Pg2rTUK8XNZ29x98VnCEpzddlz+mjDpGxoasfo12yHczAjkZ4URhResFpGykLjh57VxRWbZctyuaZoNLNFyWRisSZSGNhrphASXScB2zsbqZQ5mHLvASmxXq344ue/wJv332T/4JC9/X2ULlBqgq32qPaOuXn3A7z/z/1Fjt/7LejpMc3iGer5HTwNXTS0Y2LoPeRmICbBimMyjKPEGIQQSUGRomP0wq+L0TKMsG0HlqstV1cbiOKCvd1YtJlkGpFBZVK7NTD0HedPz+l7OQBj0ERv6XuFH41QWHxkGCPDqOi7BFhSVNz5yDc5ZcWwS3uSF0Jb6INobrVSOCO43YAXomQy7GRXwldzFIXmxs1jFEJ0jUl+jQjytdkm3hpDUTlJH9thWcnjrBU8T3MtxRGjZCNcJLGuyNvSwOi32EJyMmKMqNEQhpLNeeCf/++/z2c++Qbj1gBDdpCRrksroQOQHGBZHFn+8x//Dn4sF7ykc+FKwu17e+kgUrwQZEMmfL6M8Wl7Te3guihqQpb8qLytjTnYWba0u22bgDhapBPsrNeVymFG+efIekPakJ1lubDOCpJylNOa/XqPYXKDe1/4HfzQMS7fJEWPUZZkrBxgacxh1bJR7rtW9KP5dTdKxlLrxGXaWIdWBu8jPki3V1cldVWjiZSVxWqIQ58PMvD9QOUsYVphmzLnoHSiPXUWhxNaTdgZcAp1o2kWRBLnZ1dMD3oCoE3NZnlGHCOrq1NC/xRXBqrGYEuD0gGVetAwm1vQET+c0reJQlmawnK59YxjTTAFezcrPlQ0NPUVRwvP4fyA2ctLvvb6ks2Y6OVlzmIbhYriKIyKtFdr/u2v/z+8/vXX+OFP/Ajf8/0fZ//4EFeVlM0E7cA2hlsvfJiTu99KoSKOns3qIednX+PRvc9x+egrtMNDjooRp3qG2EOS+2QYR2JKcr2VoqgmGFWx2bS4UrHYmzLDs95s0UrRNHs89+xNjHHcv/8axhhu3rxLip7Ly4f0w5L5YsJs2kDULC/WBK8o6xoftrT9FpN8hl5U5m161HUA+bv3eNc7vUf3f/slwWgCZMqHpCtJ55fkSBJuV4AUVB7vdg6wAbQXqZeSDZUyCZW5Sco6lLESaGyzhlBJITNWxlSdRykFmekv7b3VOtvf6XxDSKeXCKB6nMsjYYCnD9f8yi//Fr/5L79CvwaQeMWEZEJoJcIwcWG2TA80P/ZTH+Nv/K3vY/9kSlQhu8CrTCQ2KDRaG4yWzaYUNPL2drfNVYRdVsduBE47Ybe0a+kduR7yudADhAWfbetTZCcIB2RplD+MSLGMZPmRUqJMyNfGaAvGUVQTUJFuc0bqL9BhTfIDhCTGriFjjsjPhYBziqpyTCYVxiL+bVZRVI6icDhnUUajHESVVQVWUZSGqrQYLdIoqxIhjPg4kgwc3jzi5t2bHN3Y5+hoRmUTBRGnFePo6QbwSYs/YlbDiL9hIKSeqihZXbZcnV6xmDjmU0NdJazNpGAldA2lxYexrBwQUFE4pjpFCIa+g/U6EFJBN3gODmccHUypC83Qbbm6koyNYVSEJK4k10UvyagZMqxCiFydPuVLf/xFHj14wKSeUJSlTEHdSPCepBS2mmLcHFscoIs9JvvPcuuFD1PUz7LpEm3qsIVnWiWqUlOWhrKwKJ1wxYxkKk5u3ebo1m0ChmEUOyqfPN3QMgy9QCLKsVm1rJcx/W1IAAAgAElEQVQrvJeAch86VuszhrGVo9IUhGgI0bJ3dMKtu3cw1tJuBwpnUXiqqqAsJKPXlgU3/sJ/88093j558KmXpJER141xHK/JrYJJiVBZmPxyUxi7s7eJJOVROsrNrgKmULKaLwy6sKjCiDuFEcBe56BknQXweX0lnmVatrLWWLSRIKIYs8IhZ15k1gggom2TDI/vX/F//NJv8m/+7z9hu4woPEn5zIkzeWmRC54y1Av4Sz/8Lfzgj32EyZ4hkLXARSkhykkE2rviH5NwmZQ2WXUh1yUCPm87E2T7nl3h213Db7S4ku4v0wt2C5Dks7WTuE8nIkG9rd+F3AFrMVON2uApiFqDCiQdiVYRnaRynT68x9XZmzBucUp85oZheAdWKgB2M3VUtRFpks0xmcT8Gukc2pSIjBK7mEL+90EOoJypoLUiJE/MWGBI8pw63+L1SEg9w7BFKwns6XvNEAtwJaY0uBqgx7pIPVPYUkEynD+9gtEzKaDQgwShO52xz11+aw7dSekahiEJRlwYmDUlYYw8fnzF6ZMVZVFh9YjRIz5UvPr6JWeXPcpOcPUMY51QraJs9qOKBJ0DxTM/c2h77r36Gl/58sv4caSsq7z1lNdIWYOyBq0E3zW2JqopPk2Z7T+Pqw5Z95FAoNCBWeVomgpXyb2htWMMEZ9GkhbdeAyJejqlKCv6vieGSPCB7WZNjD0pecaxI8SOwbcoY9g7OMbZiqHz2NJgS03XbWnXG0zyFIVn0hgm1VRctFPFihs8++f/zjd30Xv85u++FIOYSQq+JClK0uAJliVMfuk7TOYaSTBKRGmhcShj0VYSv4zTaGfQpcvbXX2NXUmHZ9l53hmjcK7II6xCKytBP6i3XSdI1+liIrCSTZdRBU/f2vCrv/xJfvM3vsJ2mSB50CPpOsAnW0vlkXayp/j4D32A/+wT306zb+nGjm3Xsl5t2Gxa+k4Io1pbyrKkKit2iVohhAziC4cxoglRxl2x6Hk72Wx3vVTuiFMueNIxynImkTtX5Qmxpx/WBL/KgUsOTIXRCk0QZxOEE6eUIapsVa4BwnVwubMVyScevfUm7eYKpxPJDyJeL6wYBxSRstJYl3CFQmym3s572D13CdT2mWgdEcMdoQXFKF29syZL0bITcLY98rFnO2zpfEs3tnkDbemHyOVqoJkfcuPuXfZuzJnMDZhBfOAqTTlp6IfE5fkKlQKVg8KIxHFxsMBVlpgEf1RmdwjKSaONcCuVTZQOSpeYTwvqyjL6xMXFVl6P1FJNSq42kYdPAx/489/GX/3xv8FHP/YfE4mcnp4ShhGVDymAuJMXoiBElk8v+NpXv8bZ2Sl7+wsOjw4py4qicNcdbPSBFAW32/Y91fSQw5P3cXj0fnRxxHZIrDcdbR+wRSG8VB2oJhUkR9dB9OI4s9xcUZSVvEeDZ39/j5ObB1Q1zBcVxydHxBRo+45JMyNGxdOzc4gWYypIGmc0znpi6jCuYIgVF0vLyu/B/D9g/+4PcHTne765Mb2ULXKMtuLkqlO+8TNpd2dJZF12Ika2QCmirJUwaJNk5QXYvIFNWt6R2liUdghWxTv+ZBb6dQyhzh/Lc9LZGw+di2sKyJs+5q+Bt+5f8Ou/8ml+81+9TLsClTy8Y0ursvvxruDVc/iu73svf+VHv51m39CNLUpbwfCSlrBnWlarFQpNWdbMZg2TZsJ0Jgan23ZkvWnxgWwGCcbmjixptHK5KGTG/zVEIhw8o7P9UGbpJzyJjk17ThgusBqsWzBZJFxepqSxlexVFA6PNu7/I+/Nni3Lqzu/z2/ae59z7phzVlZlZk1QEzOISQxmFkMjiQYkLKSWZEeHHW4/+M1PTXTYf0DbEY62HWq3NSGgxFBAFTMlEJOKQYICihqzsjIrh3sz73Cmvfdv8sP6nZOFw/abkgi8IzKoJDPvPfecvddvrfWdcLbGg/Ds5N0tyVsDjt7+EpSFJ/7h88TdM4yqOY5EiqJlzSQJZHdKbI2saEFjpLjrSsxjStK9yaEjciWxJhI02GpNTpLhu+i6oCSp1RqDaH9jB9VolYOHDzKfTRl3l6DSmMrQjBxZRYLvmc9ntPOeA6NVDq1ssL8TaPfGqJTBZELumHcJU2lclcSbzixAIL0EIyhshJgCxgS06tlYq7GV4+p+Yjq+BASakeeel25w/Lbncfs97+P2u16NtjWvfMMr+cwnP8P9n7yP7fOX0DGSslp230HuKkzO7G/v8+0vf4NzZ87y3g/+Lm98+5vRBrq+FytT30PUWFujIuzs7DObz1CpZlS9nI0bbyTMLrK//RSXdh/H6B3qwZy+A5LiwjMzSJbjxw8zsh0qZDZGI1o3Y39/WvzyJqAUbZuYtT0bawepnYQCba4cJfqaZ8/u4WrP0RNDEpreH6AbD4j5APXa7WweOo1qjtNx+LrUnede15+crIGSwK7K2KaLzRNo2eMlI+lP5aHIqDKmKRlpoDC8WY5xRluytmhdCXKrFcYgVuNOEqUglcX+IsdTtshKG/kzJXm8gUBWYpW06Dh2Lk/53Ce+w9fv/zntPmi8dEgoconwkUBuR8YyXFO86V138a73v5y1QxUBL+NLFH2q1ovMjCDee1kRYs+kSKiGwyFr6xsMhqscGR0ihMz+ZMb+ZEzfXzs4jM4YIz+XNuJRFssOLydNjtf2hWhDLl1RP9smdtso49CVxDwyvcK5Z59l6/zT7F+9QuojjshwOEJZx2BtjVO338Fg4zDVaBNdrZBtQx7UHD99B/vnn+T8ziUsHTn3Qu9J0jXn0jWTvawVrCVFJO0sLfqmvESedUGkVUxYVQ6okEkhkE1BPnPG2pKsljKNM4SsaNvAfNyRDsDq+hobXcuV3S2mYcZorZadr6pBG6Z+j3R1zomjh7nphlOcnz+FUkLmTQR63+KsRjtJuUuUqIHiahMLNKq1TBtGQ+glS2SoA6ay7O4qpjPIacbKmmP1iGfz4AWivoQd3MDpu2/hT078l9z5gru5988/yk8e+iH9rCsxibJjXfTemkRoA0/99DH+j3//v/KLh3/KW//FO9g8doR59ECkqSyuqDa6PjLvPSHDrKpZG55gMDzM5snbyf415G6L3p+ji1tM97fYmni2LsyZtpnbb7mBlSaj05R+f5fppGO6H4mpLYBZxpBw0bGyOWRUH2R/DL94fIvzlwLHblyl6o8xXD1MNTjMqLoRqsNoNyLQMAuO/Vm4rvUHfhVFTyQJ0nkhyKK1YIx0K2HhywalA4QYLColsanWTqgsLBKjjNz4Rm74hQ22sdKxaVPCTqxYG6nyAMm1oHHII1cwZTQBbSBG6ZTGOzM++/Hv8LXPP8x8L6MRIIWsMViSCoi+R0CLZg3e8dsv4k/+m3ezeWxARDzIvI/EkCEXhDJ4et8Sig19SpkYQzEqbdkf76NNxerKBsPRKs5UrAwHtF1P2/livd6LplErjLWCiloxGpB9kzi1GOxyX+lUZuQifZtJ0z0m25d5+uou21tXmOzv08+mhPEYExIoV4p5BKX56d+vUK0fZP34aY6eOM3xm25mfXOTCs98vsds2pHx2NSisxhW6ixKgBCFHGuNZLMGX2IgdRQX4gIgyR5Vlc/OkGLEz7NEC7pE1ShsVcJzrCD7xmhcZbFKqBL7WxPOp4scOLLOynCFyXTOfDpjbbTKDSduRtUVMbdUO5fZ3drhiZ89Tr/fo0KkXhdTCGUMGFEDCbxf9M8KofaUhWouu2NFxscMdrEHTDjbcuBgRWUdvU8Qdkn+Mfb2NHYIpn4NpjnNaHOF1739jdx48kY+/dF7+crnHmB/ewedIrEkqBRIRYLOU2Jve4cvf+oBHv/5o7zxnW/ljhffw+rmCjpD6ltQLYaKxtiCTnuSqmhZw9l16uYYzcYdrOmIshNS2sGsnmW7+xE/evIJntm6wtEDkVG1z2qTGLghsUv0M1kF1bXFOcuVbc+ZcxeY9mtcuKLZmTYcvemFnH7xi7jp5pMotU4OI3w2tD7QBjEtTarHmv8faG8XtjJKLbhmcI0kJgzyoDM5ptINiUwsm5ocI5kABoxeyFo0WRuUsTgjKCwqFUNKsc82VrpJYxZj7YKTVzrKdI2/JqoIcaHIUbNzZcIDn/keDz7ws+cUvFDAB+mcNImcLQnHYFXxurfeznt+71XYUWB/Ol2O2zlZSBptHE2tsCvXrNEXI2pKmfmspetbZrM5fTdha95ir1xFKyO7PVUE+srgrIXKCooahMBrg8UoSaKvqoroPTF6dHaonLGxR08988uXefbMk2xdukTOmap2qHaOms3RvSd5IHpUlhQ5VCZ1HeO9HXafOcNZ5xisb7CytsrhQxusVhprA/P9loGT/RvGkEpxUOiluSgqYquMqSGIJyUlI0pORlO0moWS40PC95FBZQqJOpGQgqS1IhlDVI6UtaDAdo3Z7pj55DxrB4bk3qN9YOvsRXI0rB87yGit5ujBA+ip5IKM96ZoBX5Qo2vhApoKkhKnbI0ixCyvrehGheIk7iY5ZawxOCfAQPayc7W2Z2XVMtuTbBil5sT2F1w6m+l6zYmbHW54kqqpOPm8W/i9P/0wJ06e4Iuf+TxPPPI4ae4xZakQFHgFUWlMhtQHHvvxo1w6f4nffNsbeOu738rKDUdQWhNTpA9znBtQ2wptIpWT1YE2FdlZvMoYM6J2h6lXnseLDr+aO1762+ztXGb70iPMdx5juvUIl7afwKUxDR7Ti2Z3XgW6pAlug9Hh2zh+6A5u3bydg0efz/rGYarKkVKiG3va3NInD1WFxuOUuLYUYdV1vX4lnd6CiyY7mUJVQVC6hbX1QkCvcyZrsSeSZbwVSq4qXLpMGWeFaKyMmBUu+FgL4rHWRkYkro2zz31NEoWnBdTIihwtk62WL376+3zpsz9kuhOLd1yQxbIyBf0ULmHGUq/Ca958G+/7g9/k4PEh875dRuCtjNap6obptGUy7ohhsYQX00wUOFfhKsdobZ1RWmV1tadr57TzjtnE006npLzwSTNoW5GqmhCF5hENmKAx2YNKGKtoSniMVoboM9MrO+w8+xTji48x3b1AO51ilJOf20dU76ky4Cw+hYLaQUp6CWQ4lah0j1YeO+mI3RZX9wxjpVAlTFqAIVC6RJYrjdaJmCN9SGIN7zK6ydTWQtQEL3sxa/RyV6ayFkOKGBms1NQDhTIeZWVPKzb6FSGJXfr6+iYbmwcwzrB79TJXts7RTcYok1ltapSt2d+9yLTbYn1zwMgZ4nSCSwGnJUdWOY1xmVxCmBRGdtHi6ClcTSW7VVPJ/YiS5DsJtyzHuJLpQmtDzp7RWkZFMMYy7WaM2yeY2gNsVescPTXCNsdo8oBDNxznDe94G0dvOsHXvvAlHvrGd9nb3kHFhMqaXknxi0qTk8LmzN6VXb78mS/wzJmnecd738HdL3kBrqkoSwPqZsjKsCk28A60IxuNMpmqrmnqVYoYHWNWObBykMOnbkaHVzHfOcv2uZ9x7qmHmO+eYb53AZTGrh9nbfMWbrzlFRw8cSfNxk0ks0FMNSopoZTlTLMeyF0m9wqiYmgtJEtKjjBY/WevOf/36/oDGTksu6qFFjRnXeD/IrkygvAppYgqCSfIXLNMUqZa0khSEl6aKlZzUFx3zbXuzhhbdkRGsgN0cURe7L6yhPsJGqqwumZy1fOlz3+PL3/+H5jszMV/rIimF6CFjMeGhKNe0bzmTbfxgT9+EzfevEnWXkgmSUa7GBOqslRVg/eZnPql/5wvaonZvEXpeSmGmZwCxL6ABrlEAgrCh3IQZWfnkzz8WUWsXlCLe6JL+F5Jx5Ii+5d3cdGwf+lZrly8TIwtSin6vif6DqsylU7id2Y0tYLsFN5n/MJ9N2WctjjEpcOGgNPQZCdbnpxJiKRQOjuW3EgULFb/kUTSYCuDHdSEXhHGXZH/CYUnx4wP4rQ7Wq1ZWauJzJf7Vm2UaKltw+5+T9+3xLjPbNaTtacLY1QV0U7hKsVwWFENapqYiXToEPE9pC7StgLxDBqDqSK6zigLMecCukmxVlq+Z0ZcQhZhNzEI6Vr20KHIvpL4OQLaSnZvDj1kRUWmYcxs61GiX0epAxy9dQ0zXGNkVogxcvs9d3Do2BGef/edfOW+B3jyZ4/hO6F4yd2lCEpJCFPOpFnLjx/6Jy5euMSbfuvNvOVdb+fwjccwlWMwGDJqRiVmpayNAFvLPam1K4VaxuekFJkGXR1h7cgm65u3cezml7J94WfsXn4SpTWbR2/n0LE7GYxuINshQTlAY7VF5yzssJxF0mgUle0IXZAURFPkpr+snLwul3ouifV6XD/6zr/NAkhItyYj5SLRiWInHoscScJGYowF7bsWq7hMjFqiaAuxuhCRF7w7Y2z5PYWUvFBmLJyOF48hRA+7V6boOOLBL/2AB+77Jjvbs2JUEsvubxHwLbu5hMMO4DfecDP/8g/fwE23HBLPMCTi0GiH1g0qV2WnKDrIFCOL98HHQCzoX6QnZ0+MPX62Q+gmGFWhckPOFh+8oMxUVPWIqhnR+0hILagOYkfsZ+g0RekZqqnl/ZzOaXchtxWXzm+xc3WbTId1kRznVDowcobGWAwRp1k6Huek6fpI20VCkMJfaUNtDDkHUBFXSbyfVproE0ZDMzAMh4aqhoyX9zqLFniR46GxODsgR8PezpR24mXvqA1GiZ/iYFBRNYDtSfRFfSdE88paUI69cUffSwaEQmIJ6yGYRrz8KqcYDO0SvYVM9BGCZr4HuzsBZWDtQEOzkqgGEW2LAgexDY45oYzQbrTOBSijuEQnYlD4HrouCmCVRXFSN7ZYkMl7mLzC6grfO67uVZy5MCAP7uaN7/ovOHLL3WQM7XjGdDon+ECKkWcef4r7Pva3fPtr32Tvyi4pSVRo0JmoEpqMzWCzqFua1SEv+I2X8r4Pf5CXvOplDFdHJfJxkeRmUNkUc1opUMZajDWS5VyQc50VRhl0yijVk9OcEMZCB7NivJuzI2klfMmF+ikKeh9DJITAYhWakxzOKgtbI+XEwcN3XdfS9ysab3OxRpLF+oJInAhlz2au6US1RM8R8/LfUpbdi+Kz6G1kpJUELqOLFKzQUiSwR9oNAQ0ETFlat0ewuqaxlq9+6SHu/8w32bsyh2RY5F7kghTLZKzJ2WIG8NLXnuJ3PvQ6jt24gY+i89TGkKLCg9hVIfbvSmmsNUUlUqOxGAal6HV0IdL7lhz3yOEypI4QLZohlVslqig3efk6RhvZt2XpomOcMd2/SDu9gkr7VMyIXWY8HbC7pdm6GNnbCxhnGI5q1tYUg0YzajocPSq1RTonnL9c6CPOSfaqcwbnJP8gRykaMWVyn8gh4Qx4UbSjsqGpHEH32KocNHrh43It5lDpXAwmhLvntENlw3jSsro2oBkY0HMinRxu2olmNQPG0PcdSgUGjZUcV6CuDbmKROXJIRE1+KxQIWKtIvqMbyPZG7oeIoEDB1YZrTv6OAVlMMYS+jkxh1LsxPUmL0f3RUcei/FFRuvEcFhCr2Mi+GsRmFqDs3JP+F5kYKPGcWg18PCj3+frn2140wf+lPVjp1EDR+M0OYhOfWVjldHGCidvPcUDn7yfc48/DaHwKCmGBRkh+GdFt9fy/Qe/y9NPnOEd73s373n/73D4hqOibU+JpJVkwCQI3pODsCmqqmJ1bQ1TSQiTs5ZsiuEoDqUqlFmjPFGy2yzuSAu3I5VFHjjeHxez3Yx2FuPENNcgayutRQd/va/rTk6++MzffUQqfF6On6moM8QrQJjuoAvJGBY8KCl418Tzz1UfKCUnuiQwGbGPQlj8xubl3g4oozWl0xMnEU1Nuwdf/cJDfOoTX2Nvew6pfCBKCMpkSl6B+IDZgeYlrz7F7//pWzhx+qDQMpKoMlKJOVTKlf9POtoYM773dH1P2yW816SkxTXEJKyJxLhN8hewdoo1HSpHrBbj0yVZ2xrqekiIUowKoRGjxW+udo5IYjpu2d+OnH1sylOPz3j2vOfq1SwndIKqThw8NGQ0qDEkrAoYHYtE79rnlqMQgpVKOAtNY6lq6WBAQzAoL84oocuELpOCwrcR7wOV05IprEGZBd1IkACtFDEE5tOe2GWcqpjPA65yHNgcok1PoiWTsNZK2LWCuq4IIdB7X/a4mcpB5RTaymcffcE8nYJK46NiPol084hKkKMQwetRzWi9lniA1AORFMSqyTiNsWopeRSEnOVkoY3kluRctM4LrqfKVJVhMGxQKhWzC7FT831mPvX41jOoKw4fWGNveoWOxIEjJ2lGhzBVg64ttnZoY5h1LcO1FU7fcjPozJUr24S+Fx1QBl1sxGQJpNEpM9kb87Of/JQnH3+SlZVVhqMRIQa8L8T3lK7RwUKkbzvJLu57JtMJk+mUvu/pQ0/SoXxeAqv0qaP1M6aTKZPJhP3dXeazGaHrmU4mtPMZC9VILEqi0Pf4rqfvOiZ7+/zouz/g7he/8tebnEzU8uaV4iVIrvzRtZvGyz5o0cstxtAsqklJmtLLP1vYEmmdC9O/yIZMLgiue464XrrNayqADEnTTeDrX/oBn/3kN9i/2iJtnYxiqFwsrUrxUgnXKF76qlt5/x+8iROnDoo3H2Kfo5QlFC1sXhZtIFN2Jz3ScWai93TzGagZmQlGz0jpWbTZR9GSMZjakWNiNt9iMpcISlevYW0to7Ou8H1xjVEV2jQMhquYQ8exgz3OPP0IT52/SDtu6HpL3wbGO4EYW0ZrkNOIo0eO040vM7lyoYQxC7iiC/KkHUUmltBFqqQQaobkNBji3BBiFL20ko7Dh0zsM81I41aNjEjq2iGkM0CUDl1BH2DWeYzVbB4YkPWMEFvIAnLFnEAFbCVofdd2su4wEpouhpiR3MnnnVQGq7BNRVXXtJPIbDdRG009BGU1q8MhCYWPHbrQXzSidCFncoyCzioZb43TWCcGp1mJdlhFATZC8GjAVZW8rpxJqZf7FCsKIKUgetppQOWMMTNGteLmYyNM/zSMz1Fv3kQ2Q0GoUyCaiGscycDpO2/m1ttPc/eL7+IL932Bs489Teq83M9S/pdgik4Zvz/jH776TZ5+/Ene9t538cZ3vI2Dx4/KntsIY4GUcVoAG9/1dF1H1TS0fctkMhF7tsrRDBoGtbiWz9oZXdvRd6IDds5Ru4rxdIxVirpuUFpCwI3WZSUlfnpb5y7z5c/dz4P3f4n3/6t/889fd55zXfei188U1UgV0MGykG5JQVJL+khxEi3jrwWEhBujgAfka12YIhYzgWtWSwKELCgPz9n9FSG/fM2Iypl+pvjGV3/IfX/7IFcvT1BJkGSK5E1WfouMw4ytNC96+a28/0Pv4ObbDoOZk00uoIoUbisVQ6goeWGWmMi5BxVFXpUyoW/JuSWkXYweY21HjlNynKBNxOpVvII+jUmmxbgZk71dur2a+Txx7OiQphniXE1MYkSalCK5BkeFaTbY3jvDtF8B1VA3FUZ7rPH4bsZ87JiPFSkYjHYkFGYpqRMOozJy+GhtEPWJULJj8CXbV5G6SJxB9hK2Y1TGKkUKgZAtPjSiKnGeEl9LCgvXHchBLJBiAkVmfaPB2JaYp8QsGcJG62KhlUghEUq+cC6HpyCrhfuXtYzRtUI3BlfV+BlMrrYoH7EDi3UNbmCJOdPOBUBySORn1lqyWitNNhFMApsWmUziym0QGZexIn8E6LyABUXXnIppXgyK0CeMgpws7awndIaV0YpwSp1HxT1UPM/40j+xeeRmqDVKOUzOqJCpjWVlVKNSYjCqeN1bXs/JW07zpc99kYe++RDTnbE4ErO0VGDp2u0Tl558ho//2V/w5C8e590feB93vfgF2NGgEMdZkqytMfTB07VtifFUEuY9a+najomeAmL5vlhDVVWFBrp5KzRGa4Vyhuw8JZPGMJ+2/PihH3Lf39zLj//hB/jJ/J+jzPx/Xte96D34pR/w+re8gNGGAi0IV86xcOjkBgGWAIdOIFbn0sVZY0gJQpAb3rmqaE1jufEl91MaQbVM/1pcC5G9tUDOzPcV3/zqP/Kpjz/I1oV9iKUI61TcipFxNiMdXq140ctu5/c//C5uef4RlOvAlDjGrFDWFurLwllE9L0hCfqq8mL/liF5VPbkPIO0DXmbkD0pzUB5fBeJ1kkIDD2YDtScwciysjIixhnDQeDQoRVyqpm3HSGOmfeeeRvY25nyvW89zNNntmnbjEkRS4+pDD4k+klCpRk6dQxsYHVlvjQlKMeHODIVhcQ10MtIupnXEDJhnohtJnlNpRw2G1zWkodkLF3IbJ3zaDdk7cgKxs5BCalaZyO7vIWo1UQOHlxhtKLoQ1sI1qKTFl/B4rlHXs7fix2wuEQ/xyvQKLKDZjBAZcf4yj5hFmgajR1WNOtDTK2Yz2ekeZAwoySuPFhFMrIHdrVBVSWywJaQnRwlE9ZUSwJ1IqGsln1hSKVZNoQ+M9mP9F3GKLGnSiFTVVpcpY2sTZyJ9P0ZLp39Jnq4weFb3ogbHkOrBq1rBoMNVkZ7zKe7pORR2nDqttP80X/9p7zgZS/l8/d+ljOPPEHuggSOyy0pLUWWrq/fnfLdLz/I2SfP8Fu/+y94/dvfwvrhg2Uy0hCl8FWuEl9FLcofW0wu0sJkAbDWLm3YtBajCaMMqtwvix1JSpHsM1cuXOJrn/8iX/rM57n89Dmyj1yzuLh+13Xf6f13f/KvP9I0jlOnj+Nq2ZVpLYvhpdElC2PMa5kOS66dMsXqXAJ9TFnmL6kOSpVdy+LfLVo14a1pnQuoEOmm8I2v/JhPfPTrbF8Yl4KXpRMr1lNyDIofnanhnhffwu9+8C0cvWmVemRKx6rKOL7wrtMsjT1ZWBhJ1yh7Si17vpJnSppC2kbrPVLaQTHHaJGsPZcq0nVz+l5yOlKSMJ3aVlilqIzC6hlxfhG/t8szT5APA8oAACAASURBVJ3nu9/8Eb94+ClycAybVazS9D7ifWI+nxOjBJZH3+LqyGDF4GrQJhVqkbzOlLSMqVEoPTkpUlBkbyBYcqcJnfyczhhMirgsYUWQCB72dnv2xz1V5VhdH1A3mqqSz0YQekvXgnGa1Q2HTxNC8PJeZoMiS+dlNNKIZpLORZVRFBQs7guFdgrjFLaReIDx1TndfkddGdzA4EaWauRIOoKJxd+vOO3EJKYOTmFqg64UzaheMgCuUaKsOAxbQ8qBrgt4D10Lvs2EPtH3idBLJ61KHIACnNUMhgZbSZpdLnI2Vfwbn710icHaJqO140xmgfFsAsB8JhxP0TN7fIgM19Z48W+8jDvuuYvxdMLFixfxXVeEawvGIwXuAxUz+1d2eOThn/HsM+c5cuQImwcPCNczS1OgioBgESWptACOzjqqql6adiyeOaWU0MEQhoS8Qil+ofM8/NA/8pf/4c/42mcfYO/iFiouPHbgv/3IR369d3pXLk25795vYUzkDW97Ac2KQ2lT1ieSY6GMIcVCTtbilLJoo/MS9QXposrUqa79HSjhQlphbEGLUygPBuQUaceBbz/4cz798W+zfWFSCp4oBQp3+lrBUwpTKV7wklv44IffzslbD9KnKTFXBd015TUmJM3tWhA3qYxfCOVBOHpKrLNiQdAC5NSgTYPSc7TSYrkFxOSJoUfpaomqGm3ouxm1ntDNLrPdzskBLGPa8R7zccfk0pz9S9vkvqa2A+aznjCPBA/eS/iONRUhJiazxLMXWoZriuGNQxnJUyjB3bm43gr5lvL+6aQgJIgWbQZUQ4OfdegYMUp2POILZqi0Za2umE48zzwyZlgf4abbj5LdHllPCVnS03KdGQw1fZ4SozgMGy0rkBzF5NQo6fDE+iqX3Zp0C9oa+Xe5+PA5g3VOitG0pzaWqjboCrJO+OjRKslrJWOclQV9H8TZ21h8lKhF1SasK1ZnWqIMZR0jzt6S35rp2sR0L1ChSpE2wjt1iJN3LqTq8r9k8J0gp8krCBnjJxCe5JHv3cvu5X02jr+Meu0YScGknRE9WGqhY1kxfrWV5Z6XvZCjx4/xhbvu4P6/vY9nHz8DXhLYUrmdxc0ooxPMruzxrS98hXNPneG33vfbvPYt/xlrBzfI9rmGtlIAhbwvrIQFxQslXLvF3ylDUdldKHKIbF/c4u++8BW++Kn7ePbJp6GXohqXbIjrf133Tu9/+rf/w0dm08TZsxcZjhw3njyMdXLipxzKWCrvmy4j7oLILLK0yCLnVelF8aPQXhQLba9eUgpiKWRBCmQWqsK3v/5T7v2rb3Px7FwcWxSghFgr39UsC56yibtfdJoP/dG7uP3u47gm4ZrFyCznp+wKryG0LAuE8JVyCfCIC91hUqQIhkxOHpIXhYPVxJjpOi/WUn0kB0H8tNGyb0oRozTB92jEFEDNd5hsn6HdO8dschUfFds7PbPOkvMK2g+ht4TYE2JLiFJUU9l4t3PRDq+tOTY2h6AyvtANJOwcxB0ml7QyxO8wW4YrBxgONpjtzlF9whT6idKyV9NGYZUWaV+fme53VFXDYK1GGQkp9yFhrMNVhpw8KcsOVltL8IrJpC+dh3Rwxihc5XCVLhOCgF3GKZTNVANNPaxQVjOfduBhMKwwtZYgdQJN4yDLfrKbB/ou0vdedoFWY2srUYlW8jisM0X188s0C1MI9r6H6V4gdVDZsoNUCmssxpTRVhW01Mh+WBWidV1Vsp9MQEjUJDQTxtPzdPMpR47chLErbF/dY955lLZYo2iamvW1TVZWVtFWM1wdcfLWmzl58ynG0ynbW9ukPhTdE/+PXd/ulas88vBPuXTxIkeOHuXAwYNi5KrLLhyWqw2ltZiXlu5PqF/SYavyDAHEzvPzH/6Yv/wPf8aXP/05dp69jAqJtNw3Xltm/Ztf904PFSAqts53fPKj38Naw2vfdDduIMmpMfYoJUv1mFgCFplc3JMTKYclnUK6Q8vCPED+phQZoyGmQE6hjC2R2GW+981f8Ld//V0uPD2Xo0+Jp92CeiwFT25KXWWef/dN/MEfv5Pb7z4GrieVApqL+/A1GVsqi/RCQPaxnOqy28vKSA3OihQ0yStxWIkaoiGrAUl7ej+V8JakSN4SQiAlsLUYjoY+EHqPVob51NOkKWkyR/W7VMZj14bQreCsRiVDbAPOj6irBq0iRntaHzHGEIMiRkXvM1uXPU883mGMZm3VgXFAj1GiTlBJhviMIubCxwuCZs53p8Qu4Lx0idpKkZN9akbrwIrTdN4w2/E8+oOLqHyQg6carPEMB5Y+Q/bFMFQlbCMHjzKWrCJtn6iHFbZGQt1VIKeMsdLppNLlNY1lsDpAGU078cQ2CEppRDNqdAEU+g5tin1VFrpTH3q51woGutBtS1pbxroyURSDgRwTudCeYqvIXjGoDLU1kCTUXSU5zJSRgzskJaqRnKisZLkE3wsYV5L4VJ8ZmI5ozrG3O+Wph2fcctfvcHBlxFaQCUZZj7UZazI6g9aWlAKudtz10hdy4uRJvvK5L/KVz3yerWeeRZcdWmLh3FKeqwiTrR0e/OwDPPXYE/zOhz7Ia978BgZrI5KSkC6xbTNEH4gxAGJ1pbV47Ym3oUIlmO2OefALX+a+j93LuceeAn+tu1uIPxfEml9Fs3fdO73/+d/9u4+IT51mNkk8feYCKys1J248JOlWheIhEi0W9DYoH5X44kkxWfxCsVRlCD1Eip7YfIvtu1ayd3ro73/Bx//i77lwZlYKXl9a+dLVQenwSsG75wY+9K/eIQXPdsuTbBmvWBBnEBrMYgxIqdhUBb+kW5AQg9SQyVHI08EHQuhRKpLSnJhb+jAlpJaUxHppUfRihOAlXEX+24ubRjfBhDk6J1xT06eK3T3LmSfntOOKPDdoMoNhJXw5DYNhQ13XtG1PjBmN+BfO2ynGZDY2RqyvD1hbHaKK+sC6TFUrrAPnoHaWyhrCvKPdncKko/Y9VomTirxHFMZ/RueE1RmbDWGemE86VlcrhiODDy0pJKIXUq9GYZWFXAEDxuMe3/dUDWgrQJOrjMR31o6oxKLeWIVpLIOVAcZo+llHTglnxfUEoqSqqYwqB6NCkWIuHMwCghQbe2MFC9UW6rqWzzoLabhvI76V8dfPI/004LRmUFks0u3qpTxS7tMYRf0g57NMKnVTiSQrC2dQJIWKkDxW92yOPL67wmS8zdrKOs6tYO0QYwdYV1G5irqqxYWZTCjms83KCnfcczenbrmZ8XTC1tY2yYdfKjaLrk+8YTM7W9s8/OOfcGVri4MHD7G+sVE6VkpTEsiheL4pGeF1IUjH1vPoT37GX/3v/5HPfeyTXDl3AR3lO0SudXclfw20olod8l/997/mdvH//iP/40eUhHxCNkzHkXPnLrG2PuD4DZsYq8vGQC0NR4UqJrK0henoIud1obcVaVZajrcpx6XJJBlCG/nhdx7nY3/+95x/alqkZYXbVIKFRBwvH4uuM7fdeZQP/dHbueslN6KcL07EcsPm0tEJcPFcqo10gILcZrz3hCAKkBBCkXbJ732M4pCcSpFWkZQ9nR/jwy7e90t1Q4yigXXVkIyh6yMxQLffk2YaHSxtD3tzw2Res7ObOHd2xnRXobIt72Wg82OqqqKpR/hebKxSyAU0kdvSGimMGxurHDi4TlMbet+CimidcFaUF5VRVBgcGvqIGntcLL1KloNGYirF3JMSTF5ZMFjm00jvAwcOD1G2J+KL04zsuGJM9H2gaxO7V1uMVQyGjuFKhau0sIgqTdaZPkVBGxE9qas13vekGLGlK9OFUyeEYuHjxSSqBitsZhnZcgZzjey+iBoIMaAwsqLoMv0skgOksJC/ZVylGTSVFAIFpnRyGbDGLqWWguzKFBODZImIi0sqmSRGMjQQwvVaFWgnF5i1kZWVoxg3BD0AXWFdJZEHWuIJ+hgJqQRtNRU3nj7J3S96IdWg5uLFi8xnU8nZfc7Iu+zAMnTTGU8++ji/+PkjNE3D0WNHsc6RYpCDNiQWWSulR2G8s8fX7/8yf/6//G/85Dvfp5/MZKUNhS/BIkxBRm1nOHbLTbzn99/P69/+rl/z8ZZMykYMO1UPqeLC2Tmf+pvvYozmN37zedjakLUSGsqCta9Z8EekwBQ9pDwgwpKXrmJxpggZVKEJXeSH33uSe//y25x78rkFTzI5hY8WC3DhUFZx252H+L0/fAP3vOxGtAuk7MgxLFFZITfLf8vrycvdknRmcWmikBHboZgSKQbQmpgDPokMLmVFyg7LCjl7QhyR4irkGeQgD5kPHDhwlOHKOm0f6f2Y8XiH/e0ZbbTM9IDdyYx5jmwcqNnfU4z3lGQqVIXA7SKVs2QFO3tjUkjUtib0EZ8iCkvOQyZjyzNP9/TtHn2vOH3qANVgRpgHdJIRyahcaBGZbA2j9VX8bqb3gSaCSqnIxIQNHkloV3Z72mBMwiaYTzL9zGJshdURrwO2VjhtiJ3ktpIjdWMYDRucy+TspR1TiWYwEGJyn2X/6ZMQf7te9pXiCSUor9HiumOg9x7flfFMg7ZiC5WKgN8n4YSmKMBF37YolWUTETLRZzSapDPK5OU6RWmFz4IGk6RwpRKQJGYEqnzdQrNXWgx1tbAW6toIDSgJoBbL62gax4bxbI1/yk5aZ+PGIXYwIll5VrxK1EZhleTV5tyWQ1mjjOLQiWO8/48/zPPuupP7PnYvP/v+j+gnc0xWvzTyLrq+3PY8/o8P82cXL/Hoz37Om9/5Do6fPCEg0qKXiNDPe84+8RRf+ez9fPurf8dkewdd9n/P7e4W42xWUK+v8KJXvYL3fPBfctdLXnid6s6161egvY2QDSnbUvg8JMf5p6bc+1ffBuAVr30+rjFYYyhHkgDgKRR5WalwgDx24j6ssoQYKJDxCkg9/NP3z/DJv/4OZ58YF9Ci51rBK6+JRcGDm5+/xvv/4HXc/dIb0VUk5wX3zi6/5yJjYzHaLvaIFD5hCEI1WeTYysgro0cKiayNuLqUMUFRkbMhp5L0lSuCvwphHz+fY3TF2nCdnB0Dt87AOq7OOq5u7RH2EybD7sQQrWVnF7pZou8ydS3JU9pommFFzJr5PONqg9eebtYL1cTVWONAOWYzzXQcuXx5j0tb+4wnBzl2tMLZCh3EpZgiAww54xUEFYhWgoQihhxlrNfFrgilJPhZ6eUDpjTEPnPloudAU6GbRF0pYvTkFK/tdkMSwrM1KC2yPGeMSM1iW4xHk6TVGdFrW+VIQB89KklxscZiSseWC3ilCq9TK5EWStcdBVggE31ajqXOaFKXCL10/MY5cg6LG1toLkZLR1hWLgKZykoiJ9mFgVBjlJZgpYSWrinJLrFxGm01ts8EnwjRM++muIHmUDXl8tUfMbuyyeHTx/Fl7jRWyc43iS6amAo5vuQfa0W10vCKN76WG07eyFc++wAP3v9Fts9dQMW8lAgsuj4FqJTZfXaLBz72KR796c9523vfzcte80pGayuQM5O9fb73jW/x5fvu5+wvHifOJVN48flC8YYov7LTHDt1I29+zzt507vewdGbblhqDK7n9Svo9FQpfJacLUoFwJOj49yTUz7xF99Ca8PLX/08dAXaKZT2sp+T5d2SgydfrVhOaaCkCZAdsRfG/09+8BQf+z+/wdnHxyS/AC0gZ4d0eOVGLAXv1O1rfOAPf5MXvvxmbKWlC0upjLMLVUfhFirZcVyTxIluWH6fl0ll8m8kdpFCR0k5otBLMERng4oWrVapTIWuR/S5oY81zXAuwdRdj3GW8XiPvatz9q72XL0SmWxnGldj6yMYZZheAW1mrK9GptOehEdry2Q2I1MTk6HrO0GHU0AZRVM1BB/pug5PsZgPNeeenWDcHtZucujgCHKPTWCSgSi7xRwCla0YDmomcQxBY5UT2VgQ3avW12gOguYqosp0IXL+6V2a9Q02bmjQLkoBDQuKj2c26VHZYobyOcQMwfdUTpODQrZggo4Pq6osn0QaVqmakBMqi8lAVglbicddH3xBTCVU3FgjUj4crqpIZFFfhIjODrIoT3IXyUnRh4CpRPOKVihTTEaRkV6X+1V+blUszkxxDSpSPyP5uVoJDYecsEp4nzmW140lpCFh2mHqCaORYrL3fTbb0zSrL0HpIU21gjENSiNB6wlyCihXQZRRPpGIOnP85pO8/08+zPPuvpPPfPTj/OIfH8ZPW66FY14rfJpMnLU8+oMfc/n8BZ567HHe9M63E3rPlz57P9/9+jeYbO8u82ufi8wuOrysFNXqiHte8RJ++/c/wF0vfSFu2Mh7zvWvete/6OWqgAfFfDObZeEjOZ49M+OTH/0Oxhhe9urbMZV8EGLuKYRQoU7Il1NKY3Ck4neXYkKjiUnxkx88ycf+/Js8/dg++f9tpEVBrlA2c+r2NX7vj17PS195K7ZZbKCFdhChSODSUrsrKK0uri3lV0yEsiMLIRO8GCmE6DHFRTh6ATsWgdU6g4627FgUKleQa1brg6wc2MDZwO7+LuPJGJc9e7sTJnuRyX5gMkko24CxpGixyhKj6HhP3Nhw4WLPbN7ju8S8Fd2qABdaFunkkgfRiomCc1hj8b7FR4VvHeefDVg75tabGw4fWKW2M5IWkw9tM8RI8q2oEHJC+YRK8meqymTxZRXRhSn72ST5cU4buj7z7Jl9tBuwdkRcV7JKZC1ri+FI4bRiOADrkqSOVVbcUzKghRtptBHdr1GEPpKCoLI1ptCESnxlL8YFlXUlwL0gj1myV0JIdKlnuDoiA13opDtP4NtE6gVQQidMDaZRuMZAUqQAtnJYo/BtV256VRp+teSSalMC3FkUOcmCloNCJJjOVPigyMYRdIXCMWwcKmV8u8321kOcWD9BMzyAcdXSxGO4MmI+l3xao0IhycvOUQPZaFY213jtW9/Izbffyv33fpqvff6L7Fy8jEp52a0tip8GVEzsPnuZL3z80/z0Rz8mxsj5J86Q5l0RLP5ywVuUsqQVh04c5+2/8x7e9t53ceSmG4pKIy2/1/W+rn/R0xnSovAFKXzoUoB6cqo489gun/jLb1DXhhe/8lZsNSiL0ywABiz5eXKZ8gAbrDKkAI8+fJZP/NU3ePKRnecUvPSckXbR4VVg4MTpEe/70Gt48StvEX/OnJbB45JlK3m2upBQU4IUdVm8pxLXKDwvsgQcxZDxvVAdyOD7IGaThcogrizgaDBagJKUMspmrE409YBhk4EeV0dsn9nZnbC9PWEyrpmOFYYBVdOgCJDn9J0h+H2O3Qg337aCHbQ8/tiEeTsioSAFnDE4baQgO4cPPb6boxRYR3G5iWU/qpjONM9ekLa1axuOHLI01VToRXmR9qvBapLV9IBFAtrJotZVWpF0BiXuv6p4tTkMKWRmV3pmBx1rmw5bF7qQ1jircVbhlKGuFFVjUTagc8b3CZQlFzME5RR9K5kLKmsIshMLoXA0dS5uNTW+uLMsVD05JvoQyg5QilPoA+ubm1htmY4nBS0HeuFLKr0wtgBda3F57gpYZUXFkmKirqriRiPjtM6IA00QuViMgo4bY+QwLbvBrCK5V1T1iJtO38zO3lUJaU8KExQXn34Yn27g+S85hWWNlIUvZ6uKzQMHsNbSzsQUwJhFdowVQA7hCh6/5SQf+td/wm13Pp97//JvePKnj4h+uDxZCyBCAzpn0qzjqZ/8XB5lmfyvrSvkSQRkj22GNc970d184A//c178qpdTrwyl+GdBysVdZwFzXL/ruhe9G28Zcf7MjBwd4FE6kIqBgKC6PcSKs4+P+eh/+ju00bzoFbegnS1BMQsN6GKnt5DBiNNuDvDzf3qaj/6nB3ny57tld9NDcXiQxqCoBXBg4IbTQ97/4dfyslffhq0FZBBXFcnDzUWicY2LlyTvAvm9qAdySXJb7PAEfVZKE6MnePHZi74nxV5yNYocPGpPryRTzWhLILOyYUC19CljG42ferKBWQdtV7O3m5jOQOsRCkfb91gCA2cZDntO3WpZO7TL4azYuqKZPDvG2HV0MgxsTe1q2tyRkAV+XTkJe2lq6XTmvcjRED1ezA1Xdx3TScvujufkDRWNDdjck0JCBVk/BJ2JBDSWComJzDmKLjMJ1UOZRXcbybqTsdEr9s61NJVm86YVXDMDFQgE3IpwwozV2FoyKLquJ0XplFQx8vR9j9EGg/Dlciijp1LE3Mv3VE6yYQuBvA+BWgn6qWKQUdiKD6Oft8zdRPZiMaKi0G18McPISe67qnJYK4BbUoYY5IBLWcbWEEX3KgaeRdmgxEkxhyhRB7ZCGUtKPUqLNjznQFM5ZjPPztUJsdinTfc8l87NubJreegfHyCo07zgN95OrmrZXeqMbSpc39B7j1MK3/el08viQg7LcbpZHfK6t7+ZU7fewmc/8Um+9eWvs799FR3TL3dxSsZdXeyOsyrjcF4AFYUCYzSbx4/wxt96K+96329zw6mbhPZaEuRI4nSeUiIun+Prd113ysps79sfOX/uMvs7PfJOiBvKoileGC7mZNjf7Tl37iLHjm1y7Pgm1pYaXYANkPFWNJeaFDI//6ez/PV//DqPPnwVER30oK6BFnr5vSqUVtxwSgreK177PGwjS2EpqrJEl6BstdzNAZBtYarLSHKt20tFpaCIkbLsj2WBnYTf5MVsU6eMShIMjgpiBpQl5dTaxMlTR6mGMO332ZvuMpmNGU/nbG1PuHSp48qWZzLuMcbifUJjWB2usLFmOHR0xk23Bmw9xZohWq+yO874XlMzoFYWIlR1g6R+aZrBiGawCtmIiN44QaizZAW3beDK9oSrV+dMJ3Oc1dRWSyZGLwRX5w1+EgltlPdRS86EFiNe6fgS6CzGBDrK7suQMRlyl5lc7VHR0dQDsg5ol7CVRinJzdVG0XtP30diQnZ5xVpMoSAAXhHmCRWEbpKQFDZtxKA2LT/H8rkmUU343kPh+yklpOvKGZyzgt4mefhDkrHaVJbVjVVsrfHZF0dmkeulKB2ctRZra0LIS91zzBnnhCLTzSNtr+hSxWD9KG6wSsqezEwI1doSYqZte9oCFMSQqOuGzfUD7FwZc/bcJU7eejurmwcEGMjlvVAwm8/Eb1DLz6+MXlJ7MmVfHSNGa9YPbnL7nXewfnCTqzs77O7uQZDCV5p2eeoUS0mbUlzz5FEKPah53otfwPv+8EO87b3v5tDxI9LlC5oinFiW+B4ZOHToxK83Ty+HH31kfWPIuWcuMd7zYtSpEktScNaFRJogGXavtly4cJljxzY5evyASLHyAhCUd98oAwEe+ck5/vo/PsgjP96WMYQepTI5LUwAUvn0HGjNDadG/O7vv5pXvf4OqoEllQ9BsjQEsU2xFFYlyotFJyeosbjQxhDLvk/+PBXL+5wCKXrJtwgBFZOE74SMSYuxT8worVE4HdG0DJrE2qajHkLnA7NZz3g6ZzLrmbWws9dx9eqcyX5PP0/kYBlVm9TGUruOo8c1mweFhDsaHiaEFZ45t8dsAoNqldo1dF2LD5Gu97JoHozo+sR00hGCEHD7LhCzom378rAKVcc52NyoGDagCUW9IF24S47oE1EVVxKVcdZgUahCzJXPWhxNVNYYVUzGIuQO2j3o9h19qzHWYisjZGRb9qtRUE2FLhw6GcY0Cp0MeC0cugihIOrGynjXe6GyaCMHV0yZrA2h0Ftc46SIWjHByGRC6IlJQkcxhqgVuRC8jTWi4dXS+YDQYVKUe9pYQeT7PpAzhChdnlGGrg10vWUeagYHbuKGW+4GaxnvbUkLUA60mBUHDh0k5oixlnk7JcQZN91wjAPrq+zsXKEPHTfcdIJ6MERh0VkCoVAwm83w3i9NOaB0b0VCtjANyFlcYk6cvInb73w+IUW2rmzTt50cVsitv6S2KDEvVSiyMawfO8Qb3vlWPvDHf8AdL/q/uHvTXsvO9Dzveqc17PEMNZAsVpM9u7spqTVFHUGBlViWpdbUaimW4gxIYOR/9P8wggRxAhtBkPhLEtiK4gSI4dhwnLTV6VZLPZBNFms4dYZ99rDWesd8eN59SAP5mmqgzwdWgSwWWWfv/axnuO/rfodu1ledYl3m8hHIF6SJiDHy4P6bP9lF7/mHf/qN114/5+Rkzvs/eiaFr0hYtgTvyJFbHwXM2XJ1eeD50wtee/2U+w/Xd/sepaoJPsFf/NkT/t5//o/51r98Rvaq7gxFE3i8GIu+qAGtee1xz9f+6Cv88q9+kaY3FHUEfSpKBYYeabhH0XHOiOwk+jrSHqUpR3mKrh5VyEn8tORASZES491YpbLoB5UG1zqhPauE1SNdM2H0lpwvaboGbZfE0pGSYzdEbsc9Y/QcRs+wixy2Eyo6OjsTbVza8PA1xfq0Wr8WZ+Sy4Pvv3rDZQdevUEYxhZEEuLZFOYPSBj8FpjHiR0/wCZRiHCdCiHXECSg8p6eOx48XzOce13iskTFOK0XbO1xrifUhIAVPKL4pyPdYV1kRpVS1v62fJNlXZW/YXkVePp+YJsNsPkNbRSbWczgVVipGf2ct1hhKzAJvELUMWSmsq1fknDkGhFN3SbkIdkxpRVZKsmwRikrTNTJ4KeqfX4PRhBwJOeFauf4f09tiOgqr5YN9JC2LqDlxjD0VL65IVkoxpNKwuVWcP/gU6/N77HdXbDa3xAjWCoNwtx+ZL3uG6YB2mvXJGh8DjTGcLHv84boWysz5vYc4O8fQgdYoK1nKMcS7wmfqxHRHAsqiOkghkWNEGc3Z/Xv8lXe+wOm9cy6vrtlsNpScK53547o7helbPvPTX+T3/8M//qi7M6p2mwpJhdN3n9ljoQ0hEELgtdfe+skueh++/yffUBreeHTOyWnPe+99yH6bKEXXwpTrcUPVzixBsly+HHj27ILHn3jAg4frWiALKmu+/51nQiuYdgAAIABJREFU/N2/86d86/98TvKg7nZ4MpqqalsrpUFpzYNHHX/47/8yv/yrX5IrbX2iC/2pCjo5dofy40cWM1Gii+xArrRH+CVZVXN+RJdMDlVvRpEcggoOzUnhfeZms6VpLF2rMepASTfk+ILsPyClAdstuff629DMeHFzw7PLZ2yGS8a0JweYhkAcCzobdIEYdsR8y+tvtpyce7Abhv2B3V7z9NnEGBq07Ugp4aMn5kTTNFhnCUGiLWOI9F2Prv7hEAKlZMnjUAWVPfNe8+jRkkeP5pycgrEe2xS0K5jWMF/0oCGMCaJIR1TRpFi9yeW4J62MvHqkygVCoQq+IQTFzcYzhsRs2dD0AguQ8VNjlZEP4bGI1ddDKzmqHJ094u6QIncEV2gnKWxN14JS9c9ZiTg1SS8XgQP4kLFNw3y5oJs1aC22NR8DJSXSFCsVSDq5UhSuaXDOyK4PoBZcpcUrnGIglUJIsNl5Yko0rebl5QW7/ZbT83NmswXTFNntpzpVeA77PSEElFK0XUOcDqg8sL+5ZHN5SaZwcvaAxs1BOTAa7UR/mIK8b40xImupq5xUH8hisatzrMrMFj1vf/ZTfPYLn8fHwPNnL4je125dkbVidn7CX/2tX+eP//Z/xE/94pdp5w05FzQVA8dxB69qJobBanNX8LTWPHzw+Ce76D178g+/oa2ErLz+5j1Wq5733n3KfpvqqHssUPqu8BUyJVleXuz54IMPefPNe6xPe0rKvPsXF/zdv/On/Kt/8Zw0KVDTR4WzKLSuKWa14J2/3vD1v/UV/q1fe4duaVDVhA7UECFTdXYf/fzjeRxHX22Kknh1XHCI7Eo0X6UmQZVUVf1FMFkheEoWW5SzmhAHtAm0XUSzI07P0eUZykyMZOh6tn7iycVT3n/+HrfjS7STN21OkINCZ4fKjhQKOXmU8bz+qOPsgUebAZUyJRouLye220yJDdELvMEaGdNTPNqLxJIVppHDfk/OkVw8xoLVhfmsYbXo6g4uc3664v6DNb0zNA5cp3C9xvWW2aLHtQ3TGIiDfKDKsUcoioxo2nyqeSeaGmdY5URGujSlNdvDRFGFfm6xHXJFLwoVRc+WUryDYOYiFrbjBTbFuohTH4VEKa0p1lAquLYoiCmTqmvGGH2HzC9oCVhHMFbGGvyYOWwjRE324iM31pGy6HKMEUSVtVau/aXmMlf4tjagtOwq0YWu08x6wcrfXN8QQuDkdEHTOrwPFerZo7F3KWtaKWKMbG/3lJiYNy37zRW3wxWpJNZnD7FuBTiMtjhrSTHXpEEZa0Vgf+zyQl3aSSdsrLp7He49vM8XfuodTs7PeHFxwXa7RTWWt7/4Of7mf/If8NU/+F0evPkQdJXj6GOzcayhGePk+6K1IoRIqH+ufj7n7PThT3bR+/DJ//ANVW/djWt48xMPWCzajwpf+VjhQ9XCVzNmk+HlxZ5nz57z6M17bG9G/qv/7H/h//5nT0kTgAed6tFCDOYiVpaR9uw1xx/8ra/w137jZ+kXDUV5igr1iqr5uEH+Lhy8jrZCaZEUtxSj6LuqcPnu1xUhywpdtu76aheYY0LljNFKnAPO0y8LphvA7ih6i+YZyo0wa8mzlk30PN9ccXO4RjWJxaqn79f0zZroFYdNYjrAsE34sZCzRanCvfuW8/MGpT1aFXR2bK4C15eBHDtK1Ch8JdZkgg9474lJUOcpBHE4NBprJcHLOVAlsph39E3DcIjcXu+ZtQtO1uesVytKmSh6opiJYiPdvKXpG7IWnJdKMlLFIlLzbLQETltN0qAsaFevhBpcY2sxs+yHQCTT947GFHQ1vZdSCTdKMmnv+qqqASt1+56TSEJyLngviWYpF/b7gWmKsq8qyEOrqHqgEitdRnzDuWTCVNjcjPiD/Hk0YlNLd5Y3hXWShuZcczfOaQ3OmRp8FTH1Qts4zbxTzDuDswWtEsGPbLcbJr8nk9hsRub9OW88+iQlGfwYefz4EVoVvI8MU4ZQsN5zuH3JFA7opmd99oiiOrRxOGNxzhLTRPATMSVZr1Sse86posSyZIDUOEgQXWU76/j05z7L57/0BQrw+Z/6Iv/ef/of8zO/9HM0MwkVN8ZirZM8jCrULxSa1klMgRE/9DRNd53k+ek5Xbd+pUXvx2BDK2gtiKRcMq63/Mpfe4cYE//tf/1PuXgyUXKDHCE+LmDOgCcHx7f+5Qv+i/CP0EbxnW9ekiYtI636WMFTEjhdaFDacP5aw9f++Bf5ta/+HM3cCtJJCcq9lEDOUw3ZkU2tdJiBnIUUfJQnyFwm4eEpH8NYqu6pUnUp1U+rhIQr4t2C1Q5rEkUPZLsHd0C7QCkT2WzxNqCVI6jMZnfL1e6SKSrafs5ytcZPit3mQIkNU0yYpsX7WwbvicHS2xktmouXnv1hzUk/I5U92US6zlLSSAoTKU/MZg3GOEKIxBhpW3v3hO67hpKk47FOE+OEIhHDRN86UggY07DZj3znL665uNhzsio8fNDRzwLG7Sl2ILuIVY7TrmPTaobnnrircpF6HMpIQrDV4tqwTpGtdMuKAWcM6JaYFjz53h5dMo/fauh7BTrIQ6WAyuIFzlm6FYX6GCBCxrbpEMQhYpUUnkaKYClJOktrRfKiTH2YqY9kHkrT2ZnQXg4JnYrY3ZzBaYVyhmTEi5tLYjGf088aXAOHw5Zpmmr+hPiwxVVUjwgUCp7oJzpnuX/ekLJIcQ6HiduNZz47vhcNu+3AbnOLcYGmmdjmkRAKtnEs8kS8+R4f/PmfMl8/5PzNr1D0CUo72tmCdVXUDYeBaRrJJdI2Df2ip2lchb/KzlyCuLQc4ihop/jCl7/E5z77aVCg+wYf/J1kTOjJgt5yjcK1DcZZwXEBw3SoYn3NfDFndXJG289fdQn6MRQ9HApLoUoIlGa2bPm3f+NnAcV/81/+E14+DZBbYLorfIJ7SigCyTv+/JtXckkKGqUCR2qLbBtSFQw1oCynDyxf+6N/g1//7Z9ntqr2omJIFVygONrY6sPa6gqyrEcNJffJlErN5jjKa+RfkmhEI8q7UrDOEbJ8sHMpxDCJ0FlpIiPZ7Ij2iu30nBI87UxD9DLGY9gPnu0hMI6KyRd8mhiGWzabgdvNQIkGf9Bk31BcRLlEDomoIRvDzX7kxTU0yxHnAkXJG9vkhCoepawksaWPqCaUTNe1gCKMAWs0TWtIaYISBNmkLH3fEY1hHA+krLi8Cby83LOaK25uLZ/59IzFsmDMnlQmaD1KB9a2Y7ZccPXejuk6YquMIeuKF68B4Bkh6maJt8PpjNKBJlt8aLj40cTZak5jDdpFIgm02PlSLPX1vzNAiY6Piv2qhxSnrQiXaxeooNrNMtqJGUtZ8cOaCkjIEfbXA3ECV3NUNNXCVsDW8bUoxfpkiTHyUMxTxnvBuitNLZayM0zKMUwRp5Bw9RRlOlEKVQolaZy2LHrDYX/Nu+9NhCninOHFxQsWSzgMA/2sJ2Bo7Zw5hW4+42L3hB9++0/oFies7v8UpSxRytJ2Hafnp7jGsr29xfsRrQvOtbR9B7l2/tMEfEQoB9m1agPtqRQqHyO2gmV1nYhSEkJ527ZyRNKKmEIlBnmMcazXa+aLBbaRfeqr/nrl4+3F0//tG0rVHYc6uiMUrjE8fushXe9474dPOeyzjLqVZFwqXFHkLdWZkf/1gie/piq8iwNlWd8z/N7f/AX+xu/+AvN1K/SWku/GjuOx4lhY74jJyP6F8lF8Y657HxmbcmX+CRpeqCq5Bh2Jbk8KizgwrCkYEwn6ltBumNyGfb5iG3cM2TPlRFSKMRcOU8RHRYoWP0DwiWGcmCaPNoau7zhZLWisOFBSza0IXi7Utsm4ZqKflxqDqXB2wbhTDNtELo6YJHA6pkSsqvi+76Rjvcu2GKXL04XGWdq2QStdFf72I3QUimGIhFiYL+fMFzNylABpjRyjlEkoC13fEUJiHOuKgIy2WlIISWgtsAgQ+U/IEFIkxnDXtTWtZnliMU0EW8B+tIaQ10mjsHeHk5JFtE6RPZbw8wQJpYpgrjiimGwj0pIYBY1eNODwY+FwO1H8EdhZ93NW8E3KGqIC3TqK1kzjSIqTIMRCRBWLwko6mlIUbTnaGXNVC/gp3oVfTyGRsuy82q6DLG6d7W6k6xqaxuKnkVJ6jFlxcvoaaMP6bEkzdyQiV1c3GNtxfu8xxrZ3treC4PQFaCpHFGMkW1dpCfiGgg+iUtBaY5TGWfl/ltdKDlLHUVbGWQEtdLMZTdcKXVtJIfRhwDrLarWkn/doJxAMoxJWL36yx1uK/UiXR/7YX6GdG37tt75MjIF/8Pf/ORcfZigOpQJis3LIgSFQSrUbKSilEUGmkqwDiqMozekDw+/+uz/Pb339l+hXEn4tJ738sf8hVYud/PzYJeQshJe74OecSPW6laogWV5oEaHmcpRE1ONFmAhpZIwbfLylaUDbhNcbPNf4cqA4yQRBGZISaUMpQpi1zgoY0knh0AbQkorV2I5GW6a9Z7sU4W3KEzEVQirk3HF1GVAq8eD+jFmvUDlx/vqMq82W8XIgeEvM0m3nJPu0MUwoJTKTkgSA6qymayQEJobIuBuhgDEa5yw5w2w2R6nMfr/jwyeZ8zNLYyyOBq2C0HRsgtmI0pkz3ZKt4vZyQlcMlUFsXCEKfPPILYwpy8EjJrJSFKW5eDpxcmY5X8wo3UTOCastOityyKhkiKN4oA0aAySVBHP1MVWsKuK3FedMhUGMQVBNVYKEUcSS8FOkeDBGMnCNVqJBU1CKwoeMbi1oy2GynJ0+wpkDu5vnlNSilGPYe2xjaTuLbsCHg3wCcyGGQk4GaMhZc7vPJNPw4O3Pk1Lh+i+/z8P7D1is4XZzxdnJmusXmcXqAWNsmC0fsD28Rzrc4OOeKcB4MHz/u9/ktU98hdPXHsjl2hh0zmAKi8UCbTSbmw2+jt8Yi3GGzvRCwQkRhez+nHMUpQgxyN9rHZRS98Hpzrvd9h0ZQWqlFOtD0nGyXtN1PWiD0hajLT+ONLRXXvS0tqjjRVVs/OSiUEryNZuZ4lf/xjukGPnv//4/5/oFVUwcxL1QGqDKUpT4eFWxoDyl6vqK0qzvGX7r61/mr//2l+kW6hhZW1HzR5aZlaV1ZcnGKG/8Uv22Nc2FUjy5+Fr4jno98dbmrD5yY1BDZnIiR4lj3Ien+HJFsECTCepA5CDQgSz+067taPoW66TQinm+JQbYDyM5yc4x5yDymBCYQgFbaNaa9sSz2MsecbgNZHqGneX925HLp4W+zzi3w5iG9b052MDLi4F8aNC0uLbFNYYQJ7naKV0tFJqYijgItMa5jkgUD6eWq6mMhxZnNdtwy9MPXmCL5tOfbjlZyOJeK0UuAW0TqZvQJXHypgRX37zwpKixCqwRYk4OmaMxWR4sRYTMAFoxbBQvnmT685auT9g2YGzBKItKjjRqpjCJHm+KEm0IVbpyJG5rmqbBT160mRhCjXFEWxIKX4XbaIFfGmUpuXIXkXHPWltF6YKP2u8HjHM8eLhicXbGZr+BADeXE9OYePOtc07Wc3ATl5sBjdjmMuWOhRezYxw9Ax3ZvYHpFIf4PjEZ+q7jyZMndFrsfyEU2q5jGHYoHSlqxLpAN19h+jNuh4nnT77P8vxnoOlFQG4NBU/IE/PZEoXm5uaG3XZL33do3aO1ppvP5KEQBeXVtq1kZFSBsakZxOM4EnNGWXMXOZlyZBoHhmHEuYaT9Sl91wMKZzuUcSjV3MVJvsqvH0PR01UkHORCq6rUo0SscRgN6/OW3/z9n6eUxH/39/4Ft5fUnV5ELGWWgpGnNZIdekTQowyLU8Vvfu0dfvsPfpHFaSMyhSokhQrUPJr7taVoLfdbfbSb1ZT4XIi5kItc53IRMXKMWcKbw5G0kupFN8milkSOgSl6fBjx9oAvI4VIIkESTpuzhvlsxnK2ZL6Y0c4sudT9xxggZ2a9Zr9PhFCX8GRIjpIUqSSKcfQry4m3GJUpac8YRlSa05pTtteeyxcHhDm44+Rs5P7DBYuTlqfvD9y8zMTQyhEH5MpnLEllSqOYhpHDIWCK5RjpN41ih+q6lqbviUF0XlpZQlbsg+bgLV0Wi12r5WGnNRSdKHicypxqR8iGm+cJvKFtBOOus4h4lRBYxSGCAiWvoTKa65cT6xctb65PwIF2I0WPdRqwlCGDF86c4nhll/eg7GXTHQi2FCDXk0WRTNtcVC14oqvLKqOczLS5IqtKLkxjEE2iVoSUGIdM0Xs2Ny+IaLbTHlWMEGt2icNmxHQJr3f45GmcsIQ7Zym6EMOeEhuaUrjZ7thcbTi5t0SbxOXFBzw8P+dkYQhhxFnD1fWHzNZDvcJvaVyk72V0ff3hinP9BsZl4rBB6zOoEaTGNvVB75nPl5QC19dXjOMkK5SulY4MKW7ONfWaK9PB3ZfRNKqjhHAXGl5Swk+eafJYazk5OZVpAI11DcZY2YgmCVZ/1VXo1e/0Pvwn3xBxai14KIoIl2RRKroBTANvffI1lIEfvfuSaahqel0TNavwGB1EwFwMSjmWZ5avfv1n+N0/+iWWp642LIZSU52kAh53eFq6RLQUunqgKFW7JeLUxEeEF5E+xJDFHhSF4BGCJ4aAigWVEjHvCXEgxSQBObbgy0hgQlFwxrCY9/S9g5LIOZJiIISBHCcMRYAHWEpxlNKQk8NPhuQbcrTE5BgHxe0ms98WUrCECcZ9ZNhmVGxo7ZwYFNChVI82llw8rvWc3dPM5pkUPT6ImLvrWpx1lCzfs5wRvWSWJX/KIu2xxt3RoZVSjIeR8TDK62ENunE085b5qmW2aDGuULSg/lGVvCvYDprOklJmu034oXbPqdRMEEWK8mM+YniL2NlygN1uwjjLcj6n6x3aZJRNZJU4jIUcdc2yFbubsOzqSqUUipbXOlMkPjKLQProN4hBTsNKybFFa00qcrgIsTBNkhucVcF1tiacafaHiZw9fhrxY8K6GbP5mrZtQE3EsicxYQw4pXBKk2u+sTFG1hU+Mu48JY9ofaBwjWHEWRFST8PEajHDmUJMExhPSRMqJ5aNowSRphTTYmf3Md19+vk5yliUSuJ70kdLp5LDg3VCC7qjz9TPpBW5i7oLV68RkSB6SK2rNU3WO6FmM7dtx+npOYvFEgHCWpx1QP3oKoVhwJhXu9N79Ta0D/7JN0TNDRhN1ppSHRCq0mtLiWQVaWeGT37qESUX3vvBC/woix5VacrSwUmHp3AsThy/8bWf4ff++JdY37NkEmAo2lGQpwu1AMpoa1DFkrO+E86qqiA9Gtnzcb+UpUOIUXRcZEUIST78KVKqIC/lkZR3hLQXmrPSWOsk5NsE+t7RzxxKSXcYJi9jVJSNos4FXSzTpNnvFTfXkWdPD3z4/p4Xz0YOexgGy9XLiedPD1w+89zeBIZt5LCNjLtCGg14Ee5qpcR10Vi0zjROUXLg7Lzl/rlitbQ0jSImT4h1T4qRUVFeFdq2pev6mnkqbgStRSg77A+EUDC6IWfF5DNJOdx8jetX+Bp3qYqmtSJNETF3RUc1htlCKMW7XSZMqmYDy0EiJiV74KwpWWGyOAYKmeglO2O3S5TU0DYLnLOY1pAVIuRNCl09ujEVsdcBxtTVBXBM8VZa12V/vVeWmnFboyyPAuucJAku5QK1ONhG03QW18iDI8fCYTthzJzXHn2a1x9/ksX5PXZhAPb0nXiOVSqopBiHzO1toG17KQwq4hpQJeH9LcuFxqqMM45xGBmHwHzWY21DqqDR5XJB8CNOg1OFaX8r9sJkudhM3H/9LTANhVgfINxJUwCccxhj7i6t2miOod8ci1y1rqEUqQg8IZVc99yJ4D3DOOBcw3p9wmw2BwQIYoyt/61EUVEe+PEC5x78ZBe9yw/+j29oLEUrQRBV8zXYutsrKB3l2KE0XWt5++1HUArvv3vBNFRz0d0Vtxa8teOv/85P8/t//G9ycr+RzFtVaghxQy4WSZg/+mgLmgaTm4qRMmhESZ/5qMOTxrN8bI+XKyg0VetRIMVJXnSVKHYg6luGdM0hbNFGLtN2PmJnEaUS03QgpUAINWt1kiwGlS05WPaHwtXVyNMntzx7suX6xcRhp5gGzWGXuN1MHPaJOGhyEOKyKRZbGkiOPCGIdC17mLZ1ooEzcoHTwHIROb+XWS4T67Vl1jdo7chZnBIpimtDIc6ClHMVZOdqIRK6cS4OnzQ+KlANSjdMAa63npfXEy+vPR9+6Hl5MdC0hlmnsHWdq5BrqmsUs5VFN4bdITMMEDzV5aKFW1g/pFRvay6Cqp887LeFly88016TR0cOhcZaSIVcgZzH9jIXJb9nFsaf1hrjHCFGYr24S/CUHDyOkFG5xsvRqqBQ2si6QUkxPF59cxZPag6Jw3aiaeacnN3HzXraZUtWnv1hQ2tka5xqrEDw4CeBRGhD9QUb/CGyXi2YtRk/7JmGSX7/KFAm13WYtsNHePT4bfaHHSpHGiXHqxQSs8UJ773/lH5xznr9ZrVWgi72I9hAdWJYIxfacRQBsxwbdNWhKqyzUiglYlAu6yHdfSYOw4BCs1qt6ft5Ff3XBz9GdoT1kKnSgfe//79z/uDnfrKvt4VCKpFSEpqESUiQCUVSz+sco5XgdotSrM8sv/37P0/0nv/pH3yT7UZYdQXx+C1Wml/7zS/y9T/6Cuf3W1JFSSlUDeZJCL0MVBHxaKw/FkDLQClP+0LNZ1BoBTEk2fNUUCilPul9IEWhv7YKsjYkPeLdLYfygm2+ImnADmSXsM7j/ShRhyVSsiZFRRglADykymDzUgiDz4Sx0ETpUryyROUIJLCSNJ9sphiwOGG4KbC9wsWMPyRi9AwT5CLWN2vlAQGW4SAOjtMzz3KV6HvFfNbyl38xsLtVxNBi68FCns4iJk8xij6STE6qGuepBBbxwgYvftSbjRcRa1F0jexH7ZcaTmYjtvIDUYKecsuC/aSlmy14//sT1089hzHjyFhtySpjtcIikAGNkGyU1kwjjFNmvNnw4XcN7bzw4BMN9x52tB1QJig19F0VQbSLy4wCpCCynYyMXBR9d4XnmEqWhQlYqELoGvZTgsAM/BAx0YApxCLSoZlT+NsNz374A66vn1OagDYTzmR610LyYHMtsgIpEDNHQOlKag6BcXcQwVZS2OqXPV3NSAX8NDCGPbkY/HBg2A0UE7GuoEjofMBfv4vawQ++/T9z/7XPYOevU7RDo1HZyY7PmOrp1bRNT9cFsZsx0ffglK1MS32nbT1y9EJKhBAZhpHgM+vVmqbppViWaseryoisE1lHVPZMu/f4wf/zJ3zuS3/7ldagV3/IKAdKGonDDh1HAT7aXkJnrMVrUdpJJycfihACi7Xiq1/7GYKf+Ef/43fZbaQL6ReZv/rrn+V3/vDnWJ8VxnBA57o3QDRqmojNEZeEllviQCaRbYezC5Tt0KbF1/i8I7IqJVl4lzuR5nHElXwLlRNaAGvYMlHUNZN/n9FcMpQDqrU4O6EtxCDXwJwUSjWkCDEIXTfEBEHhc0IHyD6ThghTxmaFzYoeOPhIHAJYEe422rKaL1l0S1IsjJNi32aGNrLdTGw3A1OcSDnSOEun2topwdXLzOl55vSsoZ8NGD3SNobDHrY3A3hD8FkEz6VgTYuqBBwp/rIKUMqhjMIUDcbgkxdlZbFy9AkJY3qCN3z7L7csTgyf+9SM1WzE+kqfKYaSxFWzPrNY07GctVx8MDLcTkxF4jZNkUJpi0LfeVCp+SWZ4jNpLEzbhv1l5uL0wP03OtanDf0iy2XTSm5HQZOqS2aaRmJOcrVWQlyRYp4IMdZGUbKQUQWVjmE7SBZHUaSxFkVdUNbgGk03s4SUSYcrpnxDsYm2N/S9JfqEsPsyBrBW4RpNzoEYi7w/krgsDodA9prGKrQR+50qihwCSSlKXQO8fPohcTwQG0iNq9iyyOQ/ZDlz7Pff5uWLb3Hv7VPZw5kJWzWuWmscTc14sSzmK0pR7PfbqrRoJVFOG/JRlF/y3Sg8jhMxJlbLNbPZQmADytRdIDUfJhC1lx0qAy+f/xll+otXXIF+DEXv5ff+L/xhw/b6BUwjXYxYY8mzFe35fVavP8at1oKoNooQ5DKbSma51nz1az9NzJl//A+/R0qZX/l3PslXv/bTrM9gCnuK0hA81raUojAF4u6aq6fvM758gd7fEpNntI7StKxOH9DMTujPHmBOzlFtR9YdJIMuhkYnQkkUXWerirxKcU+adrji0XmEPKDTDTZuUGXCOkXrFKu+YQyBKSRCVIQBUjjiwlPllAmMU6dMpwy2aIYJwkEumfMWZktNWhiut5nbfWT0CaMVMJCLorE9xrWorhAmuUBOYRIRrnEUEkp70HKp3e8S7/5wx8lpy6PHlvmiYM2BN97U7LeF508mttuOXKyM9ypXwIImBpGHGOPQqoGcCDGhKyyv6TpAkwbQaiSHRKbh5qbju99P9MuOz3yyw85G0rCF4MnBkJKmEJjNE6+9aVnOey4v4fomMOwzISliBh+lyFmHQFmVwmgwWgm4NEVSdFxNmpurA4tTeOszM/qFpWkz2iRSEXE0U0Fl2dEe+XmhRKhRjTJV1zS7UnNN7NGOKFo/q7W8LZKp6xmEQB0CRiWMqdY4r4hjYhwLZu0oWsTuUQV5LWuoUMmacUzkIjtCkhVXT4w4m3C2IQwBSpRLaC4obRi2OzF6Fy3k55gqSToy05rzeyvUOKCjI+uFHAVVhdcW2amLESYzTRNN48i5Y5oOhDgSYluv4CJezjmy32+ZpgAY+n6Gc44YE03T3XnSS93Bykzbo7JGhwk9Dbx5vnzVJejVF733/tn/itUJXTJtkRElU4hNw+H6KZsXT7j/mXew63OSli4tF08pEVMUy5Xld37/yygl+RO/94e/yMmpwvs9WRVQFq0a/FSS168pAAAgAElEQVQwRVH2Nzz97r+i3L5E73dY72WkNZZRwcXT94nZkps5/cNHvP6Fd5g/fEwyjZz3k0gnQooolTEGoh6xbNHpJU3eYvEc9reoNNKlkTM3Y7mw2E6x3Y5s9iNTLMSgmIZEGApOO1rrMCVgc6SjMNOahdU0xbEzits40ujCykJjAmamWM0sVzdwvSl4L12PKYlZaxkn6Rp0lrnNNR3OdqQQiVEIurYFXRw5zbi+GPnh9zzrk56TE4PrAqf3A5/XDcuF43vfHbndOFReUKh0jBhkNNRiydM6ioA4R6yyOKfr90njmjnTVMihECZFyT0vng186zsDU3C888UT2jkYdUAl6dLyYCgDWBXQ1tMs4OxhyzQ4NjeJ7W1gPKS6CxWdpHMOayGqQNSZ1mR0njDKkorm5joyf6F5oFekOOAaj1IB7YAcUVmha97JkQVzTLKDetRQstmSCCIZfwUUKilrFk1RhUCkqIy/EWpJLgXXGZquJY2KOAaMLszaNbvhVkZKDUXnGnlqCL4w7jOL1YJ7919jt92x3UzkaAjR4ieHUR1nZzOmsGXcj7iuBVUIfkK6cdk/+kn+expPHrek/UvUdMCae5Ssq+hf3V3VtVY0jSVGSxwnjFEYq/Fh4Ha7ZxwP9P2CUgrDuCOECWscTTPHGHE7tW0jHZ76SPJP0ZAF8qqBNG6J+wvyuH3VJejVFz07boW4iqEohU9iibHBoybFeJ15+YPvsXysiU0LWnJtjc44DYbIYtnyO1//WSiK+aLg/UCiEDJVViFUXhc8tx/8JenqBV2U7Xgp4ENCGTD1kmQLlHFgOmx5/+oF9z/7Re596vOUbk4gE4tIUSgJdMApjw63+O2HOLWDPGJjkI9EiXRlYjtGrsfA9d5z8JmsHD4ovNc0psMViw2ZpTOsO5iZiMkBkwJlzPQJmtagdcYZQ9dYVusGGstiqWiawNXVRC4C8TTqABXUOetmNEaxmM05Pb3Pi+cX5KDIwSNhmgfaZoEqZzx9/4qTE/j0Zx1dn+k6MPczWnsKiQ/eHdneBnJZYGwRXpyq8o6aEau0HKBKjkxjIsSJ2axnPp/hmp4YJ9IEcbKo4nj2wYHd7gBa88UvzlgtM6VsaRcFpowaHHk0uCHT7ArN3jObZ9ZnLTF0bLeem6vA5ipx2EdCSiQJYcBocXE0WmFVxGRN9pkPfrRlu/UsVprlScN67WiaSMkTJQfRBibxhKeS6i1f1+5E3VFJULLiyKVgjNBJoJCLdFy6Hn9i1YMqZci+5XaQIt23itlsRZgM4wGmCLYxaJ2IIWIMjKPBT5a+WeOso2kLbadIYcbVC8+wK7z5xhnL9UPGyx9xGAY++eh1UJnD+zvpyIvCGlNJJ4BJXFz9EBX+jNO3f4W+P0WrjiNW/uixFa2qYjbrQSU2twdCDMR6wNv7Pfv9UD/NuZLMZXR1rqWfdXRdJ3BW9VG+zNG/q4iUPHA4POFH7/8ZZnjvldYf+HHs9KpnL9UdilaaJGsK1GHCNLB9/pxET3v6gHT0NJpCaxWNVSSbmc87QDGNAzFm0U3FQkoKUwImRnY3Lzk8e8YiTaRpEqox4jgQz2MVKlOgBLRP5MvAxbBnurpg/danac4fkk0DXQtDRIWMmwLh5SXudkvO27o4dviUpLuImdZH5iaz8xo1aWKAGApKOdp2xv3FmnWv6cpInwdUkLStEkXSblXBOU3Ollhhi91co3tFu5oxX884uZm4urmscpkd/cyiTcM0FW6uM94PXF8/R+lA22iK7wlTQumI0l6CtA9r3v1eIMXIJ95uWJ0ErEuszwLWKU7vafa3hv0ucXNbuLoq7HaiK8ze1d2daCcNHUo7EdwqzeFwQOlU8V0FZZ0QjVPHblP48+/sWK2XfOFLM5p2QqkBvSjoMaMmgxsUuhGNn98nVAwY67CtZX3aMb1huHyRePF84DBEQkhohEwcTcGYglUShJ4Af+F5eZnp54pPvLXk/sMe5xza7UElkk8QNQYthTQlQa5X2vkxC6VwlHrIOzrnRJ4yKkKjDMVI+JFQZETEPgyB7Av3Ts4p2XF1dSBkRSxC7TZWuiyKE+1fKWxubtkeLmnbJI6Xa8/F84F5u8RZy+31NdubA3GEEgv74Vb2rVmhEAtYKZaYDUk3dIszrscd+2FLeybTlvjOq0inFimFYLhms4VECkyRkAMx1KS3HI81DG0UbWuwpmU+XzObzQXiinS51KJ3jHgoJYFObHfXDNPAvcX5qy5BP4aiV/U+Rss9x2hbicRFyBLjxOSv2WwTZ29m9HJNNEZEvjMJLlZay5iFFEvvMyHAYfDidEiJvLvl6oN36cKGmQvye1f7kNFW6MZKk6vt686ilAIctlx/79tcP/+A17/wDovHn8I0S4rTpKGwu9kzvdwwj4GiC0GE6ijVkKdMTIAzrGYNplX43UicNAvXc7J6wP2T+zQOTN6jpwmdJCzIYvEloVXGOk2IimmKNDOwbSGUkUW3oms7mDnKwhH1jt32lhwznenpFw3TmFB6ydVV5v0nG3LUNM7imoZxyBxGGNMEZU/XzrHbNd/65hUXzwOf/SstD14rdH1geZLo55kwBYaDYnfb8uzpyNXNxDhq9lvNYddDMKQq7bHWklJiPATQEWMjxiac0zRdwmqDUh3jqNhcTnz3OztWp3Mev9XQdl44c41C+4xpC8qCsiLjmHaF5CfQMlIZJ/rDdj7j6Ycjl5cHQgIwpFLQuWCTwmmNTqCrXnDymRB2XF5b7t1rmc9mNE0m4ykqolXBWg1Rxl2U7DKPePVcgRJ3rDylSTFhiiFPWaaT6mIsVsS7SjfYpjAExe7lgZAm1ucLzs7WFBUYhhskAjXjDPTLOa1zbHcHiAnbzYmDR6dMYzPDsOF2OzGMByiKw2bLdn8NKeFw2GIBy2ECnwx2tuLRJ7/A9fc9YdzLpIXYMuWr2img5mhIOPliviLHTI6R6CMxHYEe8oOxhsb1rFenzGdLsZlSYWvHPIxyPCrKqKuKZthvMbbjzU986f/PcvP/+fXKi545klW0yElSSjX7UmGdJmNEtuEPjPsD1jRE49DOiG5sZtExg5Uxwkcxlk9jxA+B7CMlBeL+gN8fcCUQtcaqIhkApWCdqW/g4/Gd+gbOFQaZ0SWhxg3X3/0mw80LVo8+Sbs4Yxp27F9+yOH2mqaNtJ0V4LNSTFMgekPbLvGTZz7vaDpDWLWU9pTlyZs0bk5OCe/3kEToewRQxpDRGVxWDBOMB0kV63qLtoI792nClkwyimIDzcrRYol+wumE05mmy7SNo28dfmx59mwEZYkoVKMp0TB6kfwY4BASfnQcflDYXAfeeKx5/FbDvfvQdZ7WKZSKWAuzheYN37K58VxfFjZXistLz3YvgALpqOuHxsLRQdY6RessfW/F1mQ6hmHJ+9+/oHEJa2Y8ertDdYGipbuyVmGbjDbVE61h3CZMyhgdSSmhnaZbOk4fKrLpGPaFYUiEmEU0rOXhoas83SCSl80msd0feP5iZDYzrE8aul7TNIbFXON0gqYSebJcaHMpNftDtHnaKFrn5J8lOfQUGUtEd4kB0/DgE2+xWK/40Q9+wGEYCJP4xperNSf3VtzevkCNCaPE4G+NwxqNjxPDENHK4Sks+wb7uujcUhIB8GFQlBzZbK7Ixd9RY3JQWCvB4H4vu26TDGnaMo17Sg5gTC1eR6Q7d4cHoY2LT3e9XlFSJHvPGEWQDwIX6FzH6fqU1XKFNSK9kbyZepQ5LkZBViBASYFxuCX7HTotXkHV+de/Xv1Or9I6CvIAzRWxLnYhw2EqTFEzX5/ibC+JXHX/Firj7chKk0WEoNrDFElTIPlISRFnOxaLU8brkXHKrDojDo2cRaVfYY4A1tWglCIgy+NxxQHGb/FPDzy9/pD25D45Gcr+Jcu5omnAuERzxDulgjIN22Hi5nZLbjynsyWP7q9wq7dI+j4hRKYycLS+gSWWKpGJBYcmBw2+QMrMlw1dC1bLlS35gGkCziRKDJydn+OcJfiR6APZB6zOOF0wquHxoxld0zP6Gbf7zOXVhmILVmv8pJhCAuRSp0rD7QZ225EPfjTw1tsNn/vcnJN7CesmjI20s8KJcZzf69neh5cXifWZYrOxXF1FbreB4JFFerA0zqJLFE2mFkmEsYG2s+xuNdMw44d/PnIyK5yuZ7iHG0oTyEiAjbKGBlUZdCIT8Qcgg1YFVSKkTLNwrFVDO3MMT7ZVSA0hZhHcloItBYuMvUpB9obDpLjaZJ5de9qucHJmefRGz9p5TAmkFDD1ASl5KDIKypK+7jXr0UNEvvXXGU0aC81izuL8AeuzFf3lBWHY0+pCJnF9dcV2ukazx2hxeKQoMpwhTviYOXt4iiqZGHbMli3aGA6HwL37J3gvEY8iCQlVtyiSr1gy8bCjJIvKhnFzYPPsJduXV4TDAZ0KsRjQcnVXFT+vlMJaU9+b0rMZq1ivl4TpQJgm4RACzmhWqyXL1QJjRQ8rHeORNpTudHqidVWQNSZlwv7A7csrNivHo1dYf+DHIU4W/jYK+UYbpdC2svKwjCFg+hX98pyUJFUrai8XNWcgyFW2qXkHCWQMiQLSDJMXuxKWbnFGHA+M04YFBqOj6Ieqh/BImir1anccWcTqJrs1TaHVGR32+GdbctbMjcWuDca1oqfy4KeRxjqavuMQR9zKcPpAs1xFJj9hVEKrRM4KHWvBLzUJTLeUOKLQxKTR2ZBiYNY6lnOH0ROERD5kshrw5ZL2xPDG6w/AzFguT5jGgdubay6ef4jKicb0zBZLXjNzlqsZu0PH9354gTKBzmmcbdnuxAqXSyNOhYrcNGnB7rbjz799ze0m8rO/cMKDNzRNNxJiRFU+33I1Ml+N3Hto2d5qLl8aXrxIXF1lbjYj46CIYUbjjDi9rKeg8ZNkhswWPRRDmQY+/AFcPF6yXmtMf0XREZwCW7XEKuMizCjYVjPuM3FUxFCIRAEkqEzMXjh7R8wW4IOXTBClJYoyJsQmq8hZk3XDMGmuBs/1mOhXgdlc6MtGWUzV7yUlKxWtNMYCRXSIR8dHzoWEOIlIUgKG7cBwfUtJE9vrS1T2zFoht/jDBoKmawqqFV9wipqQFUOIzFcL7r/xBsNuw/Zm5OADh2GidR0hBnb7AeNKRb1L96XrGBljZvIBRWLWtex2e57/6Amn/X163QmkV7ck5JBjNKQUZPVk2lrUa84FBdtaTs5PmcahOlEUi+WSk/NTiehUpc5MtcMjE6Kvu08lSholVB+dR2a646Rd8PxH7/PFV1Z95OvVF70juNEYYbmVjFaamA37ANn1tIs1HoQki7xRbdWv6VxQKeOUE7V/juhUsFmLcbtI+lYgogq0qxPibeQQPXPX4XS6S54vWXI3S8xkqgmdfCdELihiLvL7ayhBsktVkzCuJauGcZIurehIO7cYO+L0nuUaVueJUvbkIWLZY8wZylh0AZuP7X+NjVQGdAMJpimRC8yXDusSWieo3lzTKkz+f8l7s17L1uxM6xlfM5vV7D4iTuRpsjmZdmaZTNvYLsogikYCCRD8JvMjuIcLJC7gEiFRAqqAcpVlucplp9PO/vTR7R27Wc1svmZwMeY+acQ1Jy9ySaGjiLMi9t5rzTXm+MZ43+fNuDpQ847QtEu+aEsbezRnbt++otYJ7wKhCXQaOYyJ2EzgDjRdy7pfoxq5f7i3Ys8WJ5Gi85Jk36HlhFevHvjRj25pNyc83QjrlaHcnYxUGTiPhc1pYnOqxK7iQqXpHOtN5O6uMuwyJXnSGDgwIpKIcQ26BnXmEvGBu4c9v/h55b1vbdme2fepRMvQYKY7C7ioSDPDQyELzOIo2ZmPWwQflKar9BvAO8YZ5lls0VItCSVXMZKLq4RGlvDuQlVPpmXYDXzxauZk3XG+FppQFyKPeXkr0HjzgNthw305sH8sFGAuGCeVMux48ZOf0q0b6mFHcIXilBAEPytkYZpBJji/vKBtNlzf7Li/ectqbZKj2mUe5IFhOrI7VFZPWo7jkcNoNkdL/nsc0YQllS2TUmW1agk+4A+enEeePe2QfE9Jt9CYOqDMM8NxwgXLhVF0+XzUL/226oRuveLiyROm0YrZ+dUVzap//LDYKAD5FXwXtUyMcaTpOnwU0Eye3xI48PS85e6L8asrPsvj19Dp2eq7quGdFEeuypgKszS0Jxfk2DDXAiSCC/SxtUF8MPQQWg38CFALXpQYBGLAx8hQRlLJUDPRR9rtOePDDXmY6IMSPAiWVqa1mjRhackRAe+pClp0QYI/Dm+F0DSENhiRuCZOTk6o1XMcM+utR8KBTQPr05ZmY4j5lAT1heqVBGTvKKJUpzabrFCLtf45mxZuu+1wsVIl2RxKIlqLKf+9IoMp5SV64uoJsd2SwobzJ4Br2N/dMu1HojOUzzTvKWVkswlUKusTodt0+G5gPOiimDadmDpFXMXRQl3z+sU9v/zZAyfnG9qVEvsEmogtC3ShENuCj0rbw+YkMu0bHu4rNzeF27cD1B5XV4RY0JJJ6QGpK0RbtDQ43/Pm5sDnLwKbS8f6RI2g4wLizf7XOk8WT/KFNjpq45lR9kNFHbigxE7okzOHRXC0fSBNj+4Q0wyCUX3GXAx1FcSYhXTUsubVy8wqQni/JW4TcaHjtMF96c4p9Ve+XksWC4DYAmX5M4fQCNTdjsNDoYkO35tND/U4NQjCNCipmmi67yfzU2fPvD/y9tVrg0HMyaxt64CSzLbms8EcTFfz9wKqov1eIIaOlGwOmOcHdve/5ObHf8bF+3/AerWmasT7luKKZduqsupXrLoeL8vP6w3E4BA26zXvPH8HEWHV94vwXJFlTCViyozDMDCMA+IcZydnNL5Ba8GTuHv7MR/99M9Z1U/I6TdAp6dLaItWqCrMRRlTZSie5mRj/lIVilNiFEKEpvOs+kjsIrEzMscjIijGgHZ25E0iHGoiq6BJSDOkKkBE44b9/Uh1hS6a9EVUlu/FyMc2pTGckYoazto7dEkNcz6QnaWidQtBZUo7xqEg3gKhm26iWyv9xuHaQJmEKpmUDrgwmeQiBrS26Gh+zywjMBCwfNKmdTSdo5JxmJFbRMw3mhSXIA8HvC+QI1o9oYuIrFjpic1hpOE23Vj2rE6c9IFNE3AnKzQG1qcNRQfazZY8eV59viMPskgYMqWYMNV5D7nn1ScTdx80bE8LoZ/w0S7wNkPJQts5YlNZreHszDEdA7tb5eJcub2zBcmwhzJHSnHkmvAyEiRAjdTZsX8783d/c8fVO1vaVU9oB3AT4kHVYgGajaNKwHcgHWRtOByEcZoXKLfQ9EKvgg4WQtP2DTkZB28eMmVeNvkSF3y6MpdCrTNaHfcPyqefVdarjqpwsYUuznRtMY/z7KjZkvqqVsSAfNb1ZceclRDAe/Nza82oM7GwzCDRAA6usdS90ESmSfni9Z6+HVl1gSfPeppYGB7uyDpT82wh4RHm+UjwQtc2jGOyk0GBuIqUXJjnxCMd/O5mgNzSNmcMOnJzn3ny9Sva9oTHaEwEmrZhq1tubm747M0nSLX5+5eUnhCtQVj0dwC7hwcTaDsh5UyaZ8Pj1wLecXp2xuXlGU1jtlBZ6ER9f0Zsz7h5mVnzGxAMZNfHwisrwpxhzIJfbZC4ItUATaTtHG3vaWPDqm9oG09sgkkBgjckFSBRcbMVCicKroVQcDO4EfJQSFmJoSe2W4bjg1koK5a0BZiBVhYhqg3IfQy4IAsGwTaBsfVIcIzzSJrdokC3/IcQQVymXwW6bca1prQX5wnBcXPzhjhvwRdqdkYZ0ZVdCDqgEsBBEwUXAAriFq0TkKvFEAbn8epw5jZHp1s0JpJOIBfE0OOksFo11LMNN6/vSYcDw8NIzZlxcqi2TPf3NL2wPdlSWo+/qUxpRHJLIxHRgmjGu4ijZXc78+nHO772rdZQ33FAFn2ZL0JsArHxdL3S9ZWaZ07PAidnjqu9srt33N86Xn0x8/AAXi2PtaQBzZW5CHN2fPyzwtfeh/OrNeuYcd4GoOIq6gzQ2ojHRRBvGSXT5JmzkAbFebtxdUvaXqkZHzxdb3a63JvjISVBJZCKoeR1KRaPXui3h4ZPXyYKHRnP04tA14yG33LW1eWUKFoX1h6G1wJynhEsQS2nivhl8eHsVOOyFUTBgXdcXj1BdzOffn5DEU9h5vRE8Y1YXGO2LJOELF2kYbfs8+QI3lBac8r274plfFCEaVdt7FMnHuaJ9vl3ufzG75B9T9FCYDmSitB1HVdXVzjg7u0t+90RpxCdBXiVWkg1/71PsxC9zTyLWrFTB23fc3Z2xsXlJU3TLM+1metUCzVuuPrWD9gfP+Lhzc//f644/9/Hr6HoBUpVu7gKZImEfoVbbZEQLWRlHWjXka6LNDHSLKZljYJ6RxW/cO9M7Fi9x2HK+Og9q6YlzI7QJKZgpmyZIKxOEIQ0HdF5Jns7FnsJqJRlaLsgq7C5mmDbtBAEFxQJlUZMhS6u2rFaCqExv+NqHYmbI/hCrpniFPWB43GPK7estmtKdaSaydXM72mGUGTZnAmSjU9GNdU/PphH10MujjpVmsVvWocRiUKaEsrEPG1Ih8qwP5DGgc47hmlgTgNnlyvWMTIIXN8fyL6lOmEuGbyJgFMBVU9QTxDjvM1qhf/t7cw4nLHJM6HOOJ9NhOsq6jNRBBeF0Ao1Z07OPd06M+yF7VZZ9RnnCu514fDQUufefLONRUVWVgyHxEc/n/jWd09oV4Hol1B1L4h3JlaWineV1i+C3iA0q5a3LxPzwWZ14tUcP7OCzjhfkAg+KrFzpATTXNBisYyuCiEKsRVydgTvuN7vSS8bUm6Jcc3lRUt2d0u6lxF+vIs4r/jle2k6/+WSQ5ySa/4yklKcw1GWeZ/YZrnxFOfotmu+/p1TZCrcvf6MKRa6RvDOZoN2c/QWPFWX64MltNtFy6+gEKMpEBwCOSBJqXFkyDO1u+C9D/+YzeWHjFkJqQIWHv9I7opNw9WTJ3gfePvmDWWYqaXgPPglo8TgAYbQ8ogpIpaQ9Nh3XD59wsn5GT6GRznfEnhfmFMhFzh/9g+Yj6/4xf2Lr7oEffVFbz8kWLDiUwbft6y2p2hsCOsG33vCpiWuWmIbCMFbgtUyayJ4m/M4swjZ6dVTpdhMQ6EpDdJ4XGN/P7uRclQyljc61sp0nGnVgqeZEzEKTesWYWZdPJiWhSre7rrqLQej7Ra6bZmJrc2NvHd060i7yfgYqWKbPACRgHfKPBypm4oEy45NuhxvqyfPS3HFSCK1eKoW47MVSFgXI3MmdkJZIghD9pQ5WQxkgd3dnjevJo5DogkN93eJ27sjITpW5x45FeSkx19n7u8LSUfmssBVsWP8ME5EeqIEMhZ5Kc5zc5O5vSmcP2mpZcQJiFNLOnM2/8Q7xNsm07vEhkTTKn1v3XvXOfq+8Or1wO42MR5aKwheFsjnmk8+3vHxL2dOrxyusbFD8C1N7Jh0prgEvuI7RaTQdC3bM0e7Um5fFOajEhrFBQiDY54e3wfLJbZAa8HFgJ8hlQTOrg1foajNw0op7KeRj1+MFO2prqPvOnw7E1y2pZqC02r8GU0GnRW1ICLBZraL9s3S16yTr2o3fudbHo4TrltxsTmh6aEc3qLlnppkGf4vlrIsSyCVDWJUFe88bduYF3gJrKq5ItkTJdKdKZMvpLjlyZMfcHb1bUpxjONEEzqQxUboHgG+QmgbLq4ucarcvr4mjRYHCbqcppYp4mLjZMkRabuO8ydXnF1eWF4GfHmAqtUo02kuTONE1cDZk+9w+d73v9oCxK+h6A1FyLUyVyX2G1Ynp7iuIaxbQh+JfYvvIhI9Et0Ss+cWcKEdL70vtGr019mbEFbFIepN7ItthIN6fBtNtuCsi5r30HJCqpn74wNtgeCErJUqEKMskhZLyhJxtCtPDvNi1zHdUtt5fFBiUwlNJrRCv1WkMTO+E8EXS/gNWuiDUuYjpB2+X+HjCtJImmZSCVDXjGLHN5cSUvJC/vh/k6Jj482tMc+WojULx/24hFUXdtfKcF0YRyhBTD8ojosnW/pTYWwOlNXEO9/s2D5UyqEyb3qGYeZwqMSu5TAkap4pNEYdFkfTBI6HzOsXI197vyE0Jj5WtTma8Bg2puAsj8LC3Cs+Kv0q03eernE0IdDFzOu2cHMzkpPpHEutZDrS3vPLnw5848MVPk40HXhtWfeXtLGy3w9MdUA1IcFydRvX0J1AN0HsBBeV0HqOD8r+gaXwyZKTIRYWFBziKy4JuVRbJvDYzWQCpr2sCq/vBuY6cnUReOfpCd2qonmPJ9E6h6Yl1D1XigpZrOiKs7mfx+bZPgTrBGOheiEh3F4/MBzectbe0LcOGGhb07PmlJdMClmcS48FXIgxLNBPm0nnknHZnCdBBHGVHDKyveDy6vuk5hv42ONcg+ZKmQZmMt73BN/YjFIW7WoTuXhyBQo3r1+TphHNJoGpi5dWFBCPhIbY9Vw+veLs6hKJgbwIkRfyHjnPzONImRIUGwVoXNOfffOrLD/2s33VX3BSx1QqoV/RnZ8RVi2+C7SrltBFfBuR4Jd4PUf03rgWYhtQ5xJuPDLf34Iq7uwc36yo3iPFilX0ATTjPJS45JQCeQmFxiudnJK0cDweaWXJx8hKFbViVi2vQD2EtZhnNC1tvq9GIo7Lc1shrCqEhItKdbahVhU8nqCZTVRczPhyT6w94teU6JZfgXnu0DThfcJRkJrRpATxxBjNLiRKcEqdipFB6pLL6kzk3a4Cz561nGw9Ejt2g8OvzhjrTL9STjbKfhJ2cuToj6yjJzxbcdgJSSOpVuZdC7Tcvd6TSsXjzTKnjvEAX3wy8uFv9XRrR2iyvbALmQQV3GL1K87GBTEKNSuuqbQR2kZtbukFtxw3r28ODMMIxeMzpFn49KOJzz9a0xmcIBwAACAASURBVK3XqBwojDzcP9DEjq5d41xLqTPz/AB1Xm5EQr+B2iuh9zR7j8SENMJ4EOZh6bacIt6kSFGcsRUpC2RWrHovriz1ihalIBymlum1sNsPbDphu3GcnUTaRgle0blSVWxOiFDU2YZdzPpYi6Ng0YchZjRktqeRtlvz2d0XaKyUAsFbR6TVpF2PmR2/KtqVGG1DK07IJTNNhpq2ptXUAbuaaM+v2Jx/l9q9z8WTb+DdGZpb08mWSkkz0ygL2dhZ0XMm5XExcPH0KSLCm5cvGNOMlsySNmPysOhZdR1Pn7/D6dUleFmg02LjGQRqWXKAZ6RWpCjknhDPuXj2G1D0jqnSrTecXlziuohrLHmpaYOJlL3gowmPPb+atSAjNe1It2/It9f4o+WvlvsWf35F3D7B+S3OteSstgkV25TVqkj1uAa0GMs/1I7Tyyvu5S3H/Y7iDR5ZUKJaxxKcSSXOm62BIymmrwrgg1rQjas2K2o8roXqK0Uz1dmR0DtHDMqmFRomsr5h3o14d8pKI8UpJVZcr8joLASITCnmF25bj1OoFhwPqSAhEGOHFpvtuCg0vYeS2T7bsFltaOIZxDOSa5lqZR73lOMtT7dXrKc918cjL+8fqOvIk3c7fB/wq8znvzgisSNry/E2U+ZMUWGcHbl0vHiRePs6c3bZ0LQWyyk+Lx3C4wTHCorlpwuuEXwBKKz8stUMimsh9o7QVl6+zMyz5e3W7Hl4G/i7HybOrrZsykC/mSlpRwiJ0LT06xXRrRHv0TpRpoHYe1azN3x7V4hriJuG9j6zv8/MB8+0V8ZjsUwSNRuWD0JFwCm12JbcNsZGXaoiiDiafkXTrHhx/ZY8Djx/6knF468C68bwTV4UPykRT/ag3jzP3jXksTDMFfEtaRjpTyPrpiMdjpyfBYJxWAnRvLxl6TJJ9pq2bcA7oWpZYJ/y5SbVeVvQSHWkJIwZDjXy/je+hzv5HbZPvsl28wGazkHsNSq5gPNM84wbRvp+TawmU6mCZcp4x9nVE3wIvPr8c4b9PU6XXs95VttT3nn3PTYnp6j4Raao+MoSbgXHYbSNclHIZtdzBKRecnLmvsryA/w6vLdtR785xcfOWn3nluUBhGhzONySmKWFIBnJR9L4hnJ4RSw71grOW8B35Z754Z75+Aq3ekrTPyWwWtwOdtG66G2GkjJoNZlAEEp1rDanFmI9DuSqdI19AIoo3hXICylYKs5bl2DZsOZfiIukxkfbMFY1hb4TD8EZcSMIrit0QanliIwDqexw9HZYkEDxHTUUdEp458jiiK0tBkqy5DFfLNs2ezXBsl/iCrN1EsUrx+nAatuSw4BvO0LnEY1Is2HyDtWJ1kfO6Lh9m7i/PXD1fMvV1xRtMt2J8vkv9iarUMfxFlI1oGoujpdvKz//uHD1PBBDQ7sqJhoOZTHiL0Jd7Nj7SBiW8JhbW2nWcOrAtWI8vSbgnGcaYP9QKVUoqefHf1tQmXnvwzUXTzq2J5HYtvgG8CO+cXRth5eIUMn1iG8dQR1eE3GV8StH6APdpmHeKdOD57ivHPeFeRZq0kWeYV1ZFSMW2xFs8dGK4fCreIY5c0zKVKC8nUnV4eMaLnr6EICJ6CqmGa5UF3n6ztc4P7uAVPnopx9xd7cnzzNnGsHfWnyAJsZBWa9XNpej0PUdTezY7weGYcB7Y905ZxGQzv8Kaa9i8+iSlEkjg9tw9vUf8M1/+79EVs+pnJCr+Vy12FxQqmnvalWmeSI2DcEZ8t8cQ6A4XGg5u3wKOF5/LsyHPQKstqc8ffc9Ts4vlkAlRWq1WS8mVk6lMM4TVS3j2XKinUnXXAP12Vddgn4NRS/0KAH1AfURDTanIwgueouudTYwlXxkPF7jyg2NPrD1E41XxLXUYkXPx0KNE4k7huHIPLyl+Et8e4ULa8qSZeu+/DoBioNi6/VKwMWOOs1MJUFekOGuErLp0NBC1YQsyxIWgaou4S3BQ4iKeguWERUL5sHhXEFdxveekiqkQuvAzwdKmXHiid4zh0QSKEzG7sM63WRZ8TYbyoLWQJ6g1oTzFRdNsqBHpc6VWQ+gSnuRabqG1jcUbQhuTeg6DodXzGVGQuGdp1vkbWLa3bF6suLkwqN+YiqFNAtlbsm5Mh9HVt0KITBX5RefHfnwdsN64/AhE0TwLHnG6Jdh0E4UX6zY2azPjsEi0AVH7IXtmWd72tC2wjRm7u53HOaAuC2Hsecv/1Xib3+UWG8z29NK0xWavkBjtJ3VZuLstOHqEi4vlXWXiT7QNK1dSzETosVb1h7mTeWwd7QPYslyx8w0VtNmVkepSixQsrP4Sc12VEUZp8myVxz4KCQnXB+VeJ05znB10bHtWppuIkhCszJOhdev3rDbDeicSOXIauPR1FDzwDTMrNYdOQT29yPeZySaUOrJ083SLWVWa6Myq8qXAeM5j7bcwnI+ULWxEA2xfYfv/9F/wdX7f8jbvVInBbeEqddqaPtUbF4cbCyU5hnfGWNP1BZTBsuOiATOLp4gCq+/+Bzn3FLwrmwJphXRtGya7U2uqszzZDftbEFSmpbwdmdzVffVN3pffdEjRNJi/vZLsStOmWtCizfLTEpIHSnjNUFv2HZHupjwqpTZU2vgeEiowqYLOEmsYqHzM0O6Yz/MpHnC909Q15MlLBqj+UsXhBZHnrDtrY+EpmcalSkXFKFFFqCsMwyWWDtvDMBgCweHSTCcpafWJWnGLqEFka1maytZycVRUFvMILia8VLoaCg1M49KmSFPttavVcijWaBcE6gKRT0U0w86Mb2hJqEmqE7QUjmmHTklmtaBTDi/pY/v4ulh3eNiYLe/p10rz9u1FftjocyVNCaC97RdxLeZ/jRx9qThd3/ve7x48ZJXnyV2xwMvXiinJwHYs1JFul8dbR6jy+xoaEsdt1jFXOO+9F+7WGm6bCMD1+Jcz5iVjz9VdodCLomaHfvdmv2+cP1aUbEuyHeAs9D0Now8fX7kd38v8PVvCG0LXbfBBVDZ4/xsQd2tEAUaZjrxED2+FcJQqTOmn1z4yHOCeYJxSOQ5GwkngfcdoEwTtG6FiudhgP3hyPXdkYvLhq99raUPiroZojCVmf31DcyJPiinfaRpoAaPBPPOOrGc4WlM9ME615ubt+b9DZXY2M17GGZSUku1C5YZLM5TlgS7xnva7gTfXjDuweua9SpyX3eWD6PVTlIqOF2OmwswYQ4Wzh1jY4vDWhE1O2chIN5zevkUHyxMaH16SnUCau4pm2RafouqUXfSNJPnhGbLk9Fq8a8q1WCt+htxvG3Q4CzLQCq1JvPXOk+a9+icyNMDc3pg20+crE0J75ygtaESGSfHze3BBr3rFX3b40PCu8RKKl4HHg6v2V3fU8MJ2p5QaAgiy7ysoosEpgbBtw1QqVKZp6Nhl5YFgXj7r92RCji1oa8zFFbTmI/TeVBnR2owu51XRwjWCeoyZFZzmyFB8LWYJc4lai000dwglIVMjKNMyriD+VCZUfKUUY34EpBc8RGqq/jOUbMSkkMOkGTgwb+ku9oT1kdk3dL2l/imIcQ1tU6kMuOrJxTP4Vh4eDVy/1DY72A6TggB5xNnl1u++/3nfOd3LvnXf6b8+K9+yU9+duTirMerJ5CsVITlNVtKPnXxoxpxyGi6C6LJeTEjf1X6jaGixAtFIqtT4cULeH39wMNdQOUUJ+1CElFKMVGx90JwDbjA9YvAj2RiGgKxrez2e7xvOL9Yszlt0bmSdpAnBSwHeR4Frda5+5CJ8ZF2LMTsaHuh6QNxEIapIMEhEpl38Oa6EmPh+fOWdV/wTUUCHHLl+qFwsoKusaN1UI80gfFBKF4s+9GNIImi2OscPG1jg8TGGWJt2GeqVrq+IaWZEJUQPPNsYmXUHDQ1WdZGwdFvL3HuGT/96C2H/b+kie/z/HvfZ2pXjOkB8ZatLFUN9lBZCmGlJAOFhhBQWfBr3gKS5LFceM/27MKu8QUlZQVvSXVbPgBalZwyJSWk6rLYswInsmRKu2WW+hU/vnq0VBfwTSBEcCSCM99nGka0PDDlmVYmVnFmtZCSFcjZtGzTMXN/C29vLCIvrmY0CJ1AaD0ilRgKq5hJ45Hjcc98uAHfouEErR3R213MvKKB7LEZY19xA0yHiSnPBBXEe1usOPeYuYL6jLpscofgcFGoYgE1dtyQpcNxaA20neLDhM9YSIoazsfi9MzC5JylY4XgqEXwRRGfiSeRchaYjpU8wP42MzzMlNnhaCiqNE4hFmwD4/DVwRHmVzOa7lk/rUzqYVNRVvTNOeGk5/bhDTWOXKwbxmOhCSf4umd6OCLVBL1d23B3feCf/m//F//uv/893v/mKdevevb7kc9fVmJsCR7L3e0q0iY7Vor5mesSdiPh73V4ulAMg3WCKhmaQr+deP6+kLJaOppNnNjdTeS5W/ymHq0BLSaF6Fdr2r5lTo77feHt7ZZ5PnJ9c+B4SBSdCOFIxBHrijIFRDJVBkJwnKwCJ6uWEJW2L/S9s/gCp7iIbZgbiDlSdcPDved4hGnsmXPhMA5cPfeExsTjIcI4zfgFJNo1StMG+maL9qfc3U88DDNrB6ed0Kn5eENQTi96djd7dMh4H6nTZL7gCuDRmhGBfrWiKgzHGXGKaqZWJfg1MVzx8c8f+NlfvuLi3PGj+H9wenrO+tnXOPqWVJNdt0uutKoVwJIyWRzJTzSxsTm0d8uNy1GrLHhzi2UFlqGfsQ6dszm3yCNbsJDGibzg3kQFxBlXQ4XiHOBpffyqS9BXX/TW29a2n67iSNSUyPlAcEcCE6tGWbcQvRKikIqp0TVX5kNlfxN49Unh/iYuIs1souRHQXFQ00NFYdMrXVM5TAPHeSSXgVxXVF3jpaFd5mGl9TZvYEV/0pGnmXG3gzyiy8IgV6WKW/yU1XRgPUir1LCIpBdyi1K+5K0hguDRkha/sblIVOpCM3EEcSAm5ShaDaldFMHRdh5WSrMqUITTK8f4IBzvHeOxQjWZgajN98SZbESzJ46ecJ+p/mDuj6qE1TOa5ow2nrBuE4U9UQphZXrF4Nas+8CqGfmMiZ0qw054+ck1P/2bj/jB733I6w/W7B9g+2zNzXBPPBZC69m2SlAHWmw08FjSlw82TixNrNpa1I46niJGs5Ee1lW5uFD2tzPj0TMcA/M0kPKM1ojHgWa6pkFcoG9XzKnQtac4n3j9phCanrt70NKyWjf0feVwf09JK/IIc7pjyneAcBM965joWlhvPKuNo+89TePx3pYqc/bM2XH3kHn1cmR311JzS4ijHY97R+g80QFk8DBrQWYgRnxfCKcrvvbN3+a5POPhTnn76iWfXv+Cdvea01Vi0890ndJvG5iBXOmCoztpyVLJRZfil1E802SnHyeVXCq1BNpwwv218snf3HL8PMP+mo+6H9JuTvnd//A/ZXt6wUOBsrhCBIxkrVBzIpMYUYL39Jv1UtyWYBD4e3kXj49HDPzi31ygHVorw+HIeBxsnleMqI14UqkoHjQQcWzdb0DR264C4gqlzNR0xMuBPhaim2mbSt+ZHEQwvZQPDi/CeCjs3zrefCG8eKHsj7aaHSZb8TtXOQ1Kf+qpTikFXKNEKttGiIMyzjNJhXGeyGVNCCu6pkWrZ849mYIL4FyHXgTm/S1t65mnCcSWCZaYZ3O8preZUDHuEx6+bO+rWiD4OMA42EWaJuv0nLdC57xtuqj2s6Rq8YHBR6iVXAqHw4zzSmwVH6qZ6bee02dCnTumwTZ2wzgwp7zM/WzG6LLAURBXCO6IjxG3WlNpEbdhe/KEWjc87G4Yjm+pdUDqQBczXZx49qQjeuG6FNKt49NfvOW3vvOM737vCZ98vuf9765x8pTx4SWDTDRLgv3jlKZiWRIs5I1a7MYUvckiLFbRLx2GuRSk85xcCKe3E/f3iZO1Mk+AesYJypxofMOqbZhTYXf3gI8dbbNiOhamyQTLh8OMVhuap9nR+BWoMOUj05wIzdasjF6YRbi7fkv3kPGx0rSZ6AsxmDdnnGcOQ2F39OS0QkpjSxpmojfEmThP7JQuCqKJGAFXzcUTBN9FvvWtD7h89x8yynOmvOL+9Qt++C/+Fz798f/NuT7w/Lzgt0fqbkZzYdUHVEZDPNVAFodzmWk/IuLptpFaMyqO4DeUoeOXf/Wa+18kQooMtwOHVw/88l//EB+3/O4//o/ZdisGTXYgVf3yPdIl/KhOytEdiW1D400LqFW+pMnoI2uSRZi+dPRLwwc40jQw7A/M02zPt+G3zQ5tvUdDpBeHezh8leUHAP8nf/InX+kX/Pmf/jd/UtOeOt/RuCPbrnKyglVXadqKhEpZXpoQHE30pH3h4drz+kXko08Kn79OvB0KuykzjHY3jiHSRqGJlaYzbV6qdleTxWPbBkcbPQ6llpGcZrxg1OBg+sDQRZo+0K2Efi3WATQCxbJJHYqTSrfy9Ju4yDVs9ifyKxpzrRbwnWZhPCqHXWGezF7marBOrj6irSxnoVYlRGeD/ejMupPyl4XSR7N4EYSmB9dmfJfxvRLW4Fs7LqZs2bC5mo1Nqy4b5kp1yW513uN8C9IxTIXxMKKLeDSNhfFYEPGEPpDqbGuZ4lGd+M53z/HtgSI73vvGOTEUapkJXvHL/FMWB8nySVg0ZZhkCEBYCL/WQVdxqIuEuKFptmj1DEdDOnnvlpmuYx6EJnR4geCUkkdKmRc2onJ/v2d/OKBacXia0BFdRx4Tw27PeDziXSDGNU1s8eIX/FfFuRbxDeNceHs7cX9f2O8rwxQY54acW9DWHKeqqCZyHmlioG0cMRbOT3u6FtpWiA20vRCjJfkFJtq2Il2A/pTu/D2+9b0/4Bvf+T6X73ybi6fv4bqe4zQa0QfHrJUpZXvtS4POgcZ5XCjEzqNFqHPDfOx581Hl8x+NpAEoNj9Vha7tGA97Qog8ef4MHw36+YjJMox7sRtTNWcHztN1K8u8kF91e48belMwWIfnnDmmLGcmc3d3y/FwWJLi+PL/g90QG+9YBY8fBn7653/GB3/4j/7rr6b62OMr7/Tm4RrJO9ax0LeOpqmEUAitdXa1OnRWUgJJnryDh7cNL18GPv488/HrxP0RUrE3YeeV6XWmhkiVjlISp6lSnTcyyTJzcw1WoKaRVevw4hnHI+M8MtcJ163pmpXZ3wI430Gc8WVkLpPFRSYz5bto5IkoAZwjOXjMS0Wc6aCKEUBUl+I1K3kyQaz3HvFmRpfFTeF9gWhiU/HgpNJ2Jo1AHG0bwdmyRRDmWvHBjvFaMhFH2MKmdNQhsr8pjLtEmjJ5MBeKxITTW0qecKsHdPMcdRfQtGi7QYeJ6bCHSTnrOk77luRbQii83cyghZFX7Abh4srz8s09w7RnfdpxLAfyPFFt6IlU7IOllSpKECOEuKVbYFkYOGcfFHA0cc26O0eblpoih8Oe2I1sTgqnDy2ffVIYdiM+CLFtKWmwgqtCSSNzapAaCKEjBEfJasN5deSxopPN1qpAzWpeblGO+z1O7HQQNeKXYPeqgncNSCAs2kyz3Vn6nbKmpMj1y5k0JOTdyDvnG9o+IhxtBBIyTVAcN0zHideffoY8/Ijtu/8J3em/Q/XnnLz3bQu51yNleMH63X/DJz/8C24/+5w87JiOt+QhUQ/KyjU8e9rQnCWUDLqCsef1zwe++MmeaWAh9jg7Cd3t2b2+pmkiP/2LPyNsV3zwg3+Lxi3dtmOBFyxAgVqpuXDcHVj1J/SrZuEFGmig1selhj2s2NmMXBXG48iwP6LFoAqPRfLRJyworXe4PPOLH/4bfv6v/px/76ssQPw6JCvHG9oONp2n6youZjRUitgmLuKByt3DxDgU8txyf9fy2cvMR1/M3B2EmiKycD1ThemQmV5MzASG0nN6mInNSNsr55ct0TmqNwkEodKIkZCDCNTEMNww7O+R1Yazq2f0cYuPkfF4R62ZIroIjg0DLqVSU0ZLsDd2wRiBdVUsViw7PChoJKVMnsGrUHyFABl7qo+ChGqWHcEIKlHN38qjnqlSXV4WIXZ3FeeM2PE4dBYheo+sHW3vmA8Nh7uJw93MPHnKm4o/ZJrtSDxPdgw7SbTNGXPXUf0l8zjhhwdaNyEi9DHAZctqZdADiZXZv+JsE3jeV1abQu8cqXEMQ2VK5oH2Tkzd5ixjOKsjOkWd2fNQM7iLCtGZ/EdyRsqED4V2NfPsXc96K4xHYb93xvDzme1aQBPDMDIOM2l0jOMRLS2iW2oNSAZKpdSJ/TDTtR3NqsOJkouNV9zimPHeLbKLQs4OFzraDuaxoBqoVSz1qzEmmeZKGwXnWmr1DLuC5kzrA6uobDZwdrZiddKB7qkkMiOzC0RpcfMb8v7nuPV3UD2hilB9JLsTXNzwzj/4gIt3/oBf/vVf8Dd/8Zfc7V/z+uYzViXxncs1TieoDVkdRSMvPjny4idHymiYp6ymjACH1sLD9TV9GwiN5+/+xT+n7Rre+d730NCQqcsm2Dy7rtrfz0Nif7+jbVa2cPOLgkFNSgVGlRH3OMszYtD+bk+ZioXQL3kdS8uPd57gHLEUPvnrH/KTP/tTpps3X2HxscdXXvS66G2jFQ20iBOyCDmBpopUx37nuL2L3N4qb25HbveF67eVuwdFk6d3wnZtDffDBEPyvNkVZpm4mxPnN4mrM8fFecRHxwbAVTILphvrOmpRuujQDKTKfL/jbkrMp2dcXZ6ydpGEJUoJ1qF0vaNfVWJUKgnJjq5vKNixoChLAbTNoz7WvYxRezVTc8UlQx+pCC4bxaXBoXhUYLVotXAJcQquLJtpb24PhZwToQmI84j3FK3kMqM6UWPFbxzbxtFuHOOuMO0c+cEzHTLtUNB8b0LodoekiNOe9ck5fbdi3u0Y9iOH44GgntMQGEVxK8+UBlIRtieBrp9x1dF0kbsbT84ObRuigFdACj6aowXnbWvoFv+rOEQCLljimcrAcLAXTKSyPSt0a2Gahe2xcHbZ8K3f9uRUuL89Mh4qw9ExDUKeGm7eJG7vrsk1Ir5hSor3a2ppicG+n5ITpWYD0ebMcRrJKdO0lvoVQkOaoetPGNvByNDFtvQ+Qk6jvR/eRhG1OkQa0MDu3vGz40DwB549W/POB2u2V45NOyPR05x+m5Mn3wF/Rmi/TmKLF4MpuJqpYhg0DSeEixN+6x99wNd//z9nePvA//w//Y+8/cWfcvaB0PiZwXVUf0rx59wcXnMsA06XiZktyRdFQSHNA3evv2DVeZJXfvTP/xmhDVx8+B2qeJyL5GpeR0t/U7wI035gWg/069Xip9UvxfnAMnKoiw6rMh0G5uOIf7zmbdhp3CCJePE0NfPyJz/ix//y/yS/fkH6TSh6WjN5rswR8G5JS2qJDvJUGY+O3V3k5nXls9cz13vh+h5ub6FR5emJ8M2vbXh+boLhl3eVX36x59Vd4fOjsp9gfyocD4WaGtqmRUi0KyF2LQjMOUEupKSkpFBNLBzVUw8T+8Mr0ttrnj5d00bHfl9wvrI9iYTGJAzaJKqAVCVPyZYny0zE7oYm11C1eMo8C9Ro4StajZwh9quUSkqFNCs+Fo4H4XBIrNae2EZ8rPhYrCMU816ajUdAKzFECmppaHUxuLdCXNmFHFeVbuvR5JkHIxSnIXF8U5EixG1i3XQUEhNKDQ7WG3KGnoKbYNxNHPcjftMStp5hrVxetUhItsTZeuJmw+sXR+4PiZOuZe1PiU7J9QH8iEghNsEQ/2oUXqcONKA6oTIZmn+RtDhvwd0EjwvQ9JVuWxiHI6uTzHiEefQMh0qelMsnDfu9MQclOB72E/cPtwwHD6mjppai0MQIFKZpYhznBfTa45yn6mQ34uIRF4lNxGkllcQ4WcRi8GaVDG2kVkcqCfERHzqcy5SifPGy8MXbW06ewz/8j77F83e/zfu//R+wWn0T1Z7kTqlEcLYEcKXFV4/zI1US2TX4bksTV4jrcJsTXu0OfHybuTxv6E6f8Ozbf8x77/9jnj7/W/7Jf/c/cPz8NWHGQKNWdey6EOXhcKR584rnvWf/aeKv/mnh96Nw+sGHzIu/GLcsKjA1QJknDg/3JuFpI05spuyWZZ0A1CWYaLTn1nlmOThTFi0ezgC/rSauf/53/PU/+18ZXnxG2e158+b1V12Cvvqi9/hyaRVSWbJEh4qrnuMd3L6cmSdh/5CZZs/DPRzulKbCN99d8Ye/c8WTbUbKPaA8f+eED97b8hc/uuFnn04c75R7hE307HeZ3ZvMvEucPoH1hVAXKKUr8iUkFEw4W4uJKJ0q5ZiYH5T12Yp1E5nKQKUgwVNCJraCRCBUsswszM/lH5Qvj6GWfeFN1T87qA3KRE2PgB7DD4kXYjWuWfR2Zyw5ghTG+UBoMk0H7ZLmVjFQpKqQs22WNdsxZbVe0fcNlZl5HvCx4jbBSLqT4reQR09OcNiNrFUIrTNKx5x4GEdi2zGmkUqi6XtknxFVdvdHmuK4b+Hd9zfUOjOVAY0t/bOObt7w05/smB8KXYlcbs7YrFrWJxOxhbb1xDbjwkRwQpQNvgouHk0epC0iwX7GEgClpkBOhVyTuTpCoulHQizkVaXfJspcKAmukkk7DCLqmbOlvh3ulcN94fr1xP6ukKZlHNBEwAgo89SYy0FHJCSDTFShDR11KOQquBAJ0eODjRYMSeZw7ZpmvULrgba/pFTlOB85f+ddnrz/R7z7rT9C2u8y6zmPDDuPyauQSnUOVyM+g8hEEUOp4YTd4ch+Slx++Mesn3d88OHXef873+Xqwz8idqc8e//3mXPP//7f/vfkl69parabsSgJJanZya7vHoibjnNfufvox/zlP5n5w//sv6L92ntkr2StS9EzlYIDpsORsetYuY35yIWFOWl2tv+HvDf7sSQ9z/x+77dFxFly66rqql7YTYqkhiK1keJI8s0TTQAAIABJREFUFDmUKI1GI0oUxpJlweOx/y5f2HcG7IHHgA0IggFJpEdDSjTM4SKSaoriol5rycrtrBHxbb5442RTwNyqeMEDFKqR6Mo8eU6cL97leX5PzUqd3m+3DLu9eo5BSS2mUI3B2UIjI1c//D5f/7M/Yf3Wm9CPXFxecnHzE5CR4azVfAogjZmMztdyVvBhZyzGC7PWMdtbjnxhecfw4v2OD72/4Wy2IcU1zVJnaOPwlJfvLTk6Oub+nYE3H+6pNTMPjlmTaB00JiApEXcFLKQJeWOrwzs1YAu6PU253NrIxpQoNWPtBBlwhewyTasggGp03idm2mCh6W611GkO56jFsdtmdtsCSYV6RrQVNROtXqg0nafxluAnTh3Tc7Fq/HdiFbedIj7oPE/Eqq91qu6CtyoLyZE0lGlr7YBCKYlIwswMs9mUIhZh7FXycnW+4/p8Q2hmXGx2+MWe+VnL8rjFeWhmc04GWO0GxjRiSiQPQp17CjuqGTGNsHjBcLRyvL0OfO97a/rLG+bec3rSMF94jFFyTNMpndiJajJDFwiNw9mOlBIpQkpVvb85ErOwG3va1uNtwFmvYwo/YNyerhOamYaSi/SHcoeIYXYmDDvDbm04vgNXTyurK8Nu7ej3I2PcUZMnJshjQKTF+ITYniIDQ+rBCMt5qxYtC2PcElPEBc/8aM58ccZYBRw898Ip919+gfe+/wO8/N7n6U4WjOV5zNjStVlTm8v03ssBxJDJUjBFEIL6aaegncXZGX/83/8PdMHrjcJ6HBZTZhjTYk5n/Mq/+WM21z3/7//+H8jn57hhxFAwFFxVYMI4Jh49ucA0joUUnv7ge3z1L/6Mj/6rf83y/gtsijqlTNUOxVCpObPf7XCNR7zT7znFLJSUKSlRxsR+t4eirXGVOomVtVJfWGH9xut87fN/zvWb/wBDZHW94fGTS9KYn+Hpo49nfujlMVKNVyrDQVNmCwaPd47UJoZVweTESRDuv3/B/fvHHM9GpFxhXdSZWpMQ0AH9uOM0jHziZ0/4yPtPePRoxW63xpZILZXQWA0QyomS1cqk8wmoqZJHRbeLmVLOalEOmpucDqLOEXF1chAYbLCIn7a5xqhVY7pDHojJGqnnWa8T65VgClgx00zL4YqjCYKYBCVifiRJy3mLDRnfVIKxhM5gQ0SsKvzjqNBK9XgdDm5wXhAp1FxVSOUmvpmzIAkkIUx4rABh7hjGimkF1xbGTc8d59jEgd1m4Oi5O1gvHDfCvFTmvSNmlRXVEpEaMNIR8wAU5m3Diy97KJW4D7y+XVNpaJrnGXtY3SR2ux3jMKoFsYyU7GlCZLaA4Pc6WypwtHyOe/fegwuOrrOIS8QRHj49Z3N9SSEyP2rJtXDnjuNoWTg+8cwXFh8S2AxFsMFM3MM8RUQKi+PMdmUYhsA4FIa+Z7MaifsWyoLYB4xrFZLgKjZA21hqhjRqLkbTgbWZbDLJwAc+9DP83Mc+wsvve5Fm0WC8o5KpxjAODpE11nSE0NzOZd+V+upWXr+oAE5Exd0mBOZnQVl0VKKI6uzSgIwdJrSEu/f49L/9t9xc3/DNP/lTuFxj4l5jFAzKtiswrAcuzm9w3jMTz6O/e42ve8/P/8vfoj29wyj2tj1VmHylH3rC2NF6RxwVdAFQcsYawzAODFFlTV5Es6WrMjAbI+wevcM3vvAXXPzgB9ixst2OPH18zbAdcc/+zPsxbG8rpJgRvR4xE1xRxDD2mXEfKUNmbmC5iLTtFt9v1L3QJJwrBF/0QyBqdo4lUlIm9Re4Hu7OhN7qbKUMI8POItbjOqPIKZzKSEi6tKuarVER8JaUlXrbdFYx8VWRTshB+iFUq/7Rg/Pi3dW8qjQ11CdRKaRUiclSBt0S2mkWV/LAfO44PjXMOvAu471FXKaaEbFK/w0ziwsGF7z+8VqdpSEz9GpAlyls2RuLc1alIGXUQbPIdKBPCPpakInbBkU1Xw10CyHuKnl0xOzY5whyg5hG6R7BctRqNkkfR2K8Yb/tsG0DzGl8C8Vy51jg+QH2A74uKasOjxJUduOA2AYXFvS7kWFfqMVSy4JaE37ulGSdK/uN50pG2hZi2RIp9DmTsqXt7rEfB27WlaY54+FbAz/YnjOb7XjwouPkdMFsWTHBYmxLyZGaIKce5yOz5UDoYNwrfTpnT7rbsO8d223PfptZ3egBR6x4B6NkUtKt9nze0XaZdpbxxy2/+M8/zsc/8ZvMju8yAmMeSHWglhGKEocZvPpOqQTfKMCiynT4TeSeQ3s46ecOguDMJPeZPkQiwmASeVzRSMGHlsWLz/OZ/+6/ZXVxzQ/+05cIq4RNFVOV/uOxkCrbyy0r39CagC3Cw+98B9M2/Nxv/Cbt8vQ2AzpPLqeYEvv9Ht8002E8lalVmXz7/Z6Ukj4/I3gxyCQT66/O+fr/83mefO972H4k94mbJyv66z0+GdztL/zsHj+GYCChZIPD0PiAsY5hiPR9IvWFPIDFMgsgVckp3iu6KTSV0IIPgneakdE0KhimZsiTlKCAJA0MAsgj7DcJHw2+1cChQiKlUeEDU6UEBdHQMazI5J3VzISK2sOsFVxA4aGiavNSlBhhxFAn1HzJotjwAi4YuoVhWyCPhtgXrBiCKzhXaRohNELXWUwj1FBxTcHPCs1cw6td0GxVax3GOqQRYhtxrWUcInGIlFQoFFLJSo82BmP1uZeaoApWLNZYKuldETGoh9NXbKORkyEbfCr0OZHrCOLJxdDNZjjn8bFFfCabQhoL1s8J/oQUE972zJvKe15YcGc25/JRZXWxB2fwjSGlzH6bsd5OOkNIcU/OnvU24UwhBItnYBguKNmy2qzpx0Q1Ft91+ONj5vMjcIFxzBwtHyBxyebqnO88vcGEHtcWjk6XLJcz3n7jIZRE01gWR57KSM2ZcV/p+4Fu1rJYzmgXDSd3FYJxdRF5/GhFvysIVoPQbaRdOPw8IsGyvDfn7oun3H/V051lxBckNeo2KeOtHU+opBwZxv1tjKL3ARG9Rv/RR/9wI53+3UEEfHvk3bojCmPZk/qRmTki+BnPf+D9/Ma/+2MuL5/w9Gtfp9snTNI2mWJwBcq+cP3khsY1nN45RdY9b3ztmxjf8PO/8RlMmINYXLHEFCkU9vs9oWlo20bJKFVpOvv9nn3fK0RCBBe8YuFcYVxd8fXPf4G3v/kazX6EAa7P19ycr7CDSr1u7TvP8PHs29uo7VYVoYyVcewZx0SKWutbsRpj5wtt61h0au8xLmEChMbgvOaxCuC9UhtE9G5cOkscBLc37PaVWnRDPO4zY5+w28wQEqFxlKr8tFI1x6DWzGEJUYxm86YcKWUEW3DW0AQdYivbTFuRWxN1rRM9hgO2GWMcoS0c35sjLvH04Q1GBO8NTVOYL9CKoQPbKC3FNhU/qzQLIcwE15SJsKHJV9Y1uqEtFtc63Dgy7A1pzKQhEWPCGINzHijTUF83yzIFrGv3WzlMrA1acVgrSm4RwReLHwtjLPo6YBB6nFFysTWVZCJj0Sw6q8AtSq10M0+DpVH2FcYkQuNpgihbriTu3HkOIzP2+8TFxTX9NnKzGfEWumIQO2B9A8WQ807Ju9ViqsfWQhpHdqsVwxgpsy3eBebtjMubgc2mx7ctNzeW46Xh4ZuW1BdCEOZH4P2cmh1j7xj6QHdksWFHLFfcf/GI5+91vPiewAsvnrLf9Qx9wTgLrnLn/jHHp2e89dYFlxdPsY+Fh6+/xcm9t3j+xRPAYkXIRahJj7OkijWNZjw4Ggx4F5hMPApYkMOqT98Pw2Fj/6MuCACdm2lQUGYYNloE+I73/vyH+bU/+Bx/+uQh6Y23aGrFlaJh86ZCFvJQePjwHIzl+OwYu+l58ytfp206fuYTn8I1M+LExaMKKSX22y2tDxxEsjVldpstOSXVsIp+tp01pO2av/vSl3n8zdcIu0gZM1dPr3n06Cl1nBYithKefaH37A+92Ks8ZKyVNPbEIeO9ZmMaq21fO7M0LYRZIoRMmDRSBaUTWxt0JoLitXMuWF/w3pJTxrQFaQXXOsbesNslXTKkSsmw2ow4nzTH1jAdenphVZSQ7MMEC0WBptY6GteQx8LYV1p/YMUd3kCDHDyvItSsxOIYCwOF6rXVwihHDptxoeCDpqs1Mweu4HykmYOfC65T/Z4LjrZtsKbRRHvRqaE48OKQINjGksfEuO0ZtlGXAanoIeZkYvgVXZII6gFD79YVpmSzyTKGkGuenmulcdryx1KIZa+eyuwQAdc4Ou8okhnjNWmfaFxLblqwBmzP0iScFzZtpescIbQ0fiR4dCnTjPS54kJDTOqfLYOA0ayUYUykpLnHCJQ0kvotKQMpM6539Dc33LtzxvPPP8c+7fGzGTa0lAJxgOX8jF52WFsY+0TqLc60ONOSbGG92jKWLeM4cPX0MU/veZ67G7hzJ3B0FOkeWBbHluKE+fGWk7OWuy+c8Ddf3/D2m28we+OMX/jYjCBOPbeSNYu2auBRmbiKUNntdqSkQvP53OKdV6nT7YTv3ZPg8BVz6CCmcKDb6k8U4zXmSNxdk9weUwsf+eWPcf3mb/PF/+0/kJ9c4lNW8ouDYdpsl5iJT8657+DsuVPqZss3v/xlXDvjQx//BNZ6rcZygZwY9luGtqHpPFTo9wPjfjM9QdWKOmOQOPDaV77Ca1/5Cs2uJ9XM0+trnjy9YMhF6TW10AiYn4SZ3rBTw36KhZynLSoQGotzGbGJ0GWaeaWdiy45jCKXjFFi7LtjVn3Dnbf4oG1mdRmXKyYYsCOIuaVQJBxxKIqoGhJmn/F+in2UwwFYyKYgLjMrXhFHYmhcwImlCR7vEkbitLmdDjwr0xYaxSeJIFLJYiayrFaCrrGEWml80SVFo4lcBW3jQye0c4Ptss5FnMfagA55Khi9wOvB9+g0MY5ccKEgRXA2MPaFfpuIo+CcVsfWTZJ6oxWtPuqtF5YJN1TShIAyh5YqYa3RwHOxStMtox6OY6LkBGYEdPtOMUQ8xUDoLF3JOCvMZg3bq8xsZljNDPttotQ1RvaI6WnnnlMzZ7+NxD7Sp0JdKwmkpkhrzRRmXdltN2r1Q/AUqhiGXc95fcJi2ZAQjk+WdG3LzeWabVPZejvpJy1xzFhfyKwwjeF0tiSXjrGP7LYbnryzYXO9Zty2vPLKkq6JxHTB4qjSzTKh3dMtlnz8V08J3/Asl0ec3X2AyGwCy06vmbV6XRa9bkvRfItSKsZslRDUzbFWP4pq6J+qcODQ/8kB33T7//yIF1YvCxVbx4TkQrds+OTnfoerx4/5mz/9c+zlGp8STYVoKn3NFIS4HYjvnIO3nB7PKdc3fPM/fYmuW/Ken/0ForW60Kjqr+77G0I4odZK319TagTbIM7qPLkUvvvVr/LtL/01dbWi1srFzYq3Hz5l2I9AxdZKS6XJgv9JOPR2m6xEVtEMAucdbafJWr4patKeQ2hFqy0RPYyM6oKMmInVFaloqItYo4jyqeLSfGxLKUk/tKLC59GmKandUsdKGitxmDJbp6pCjOjWNoOtFlMrpQj7fsTOyyTOTJO1UXFJ4pg2sjqzrNVNI4+s0YZjggyN93SdQ2JP23l1pEzGb+ugnQnN3GCcfs1Yo1tcq5qwkorO6ZxlTHEi5WrmqZgMNamdTaDzltB4tqtIGlW75rLDFTA2aVU9HaJQtcqdhIOl1OlDOd0Qqs47rejW22ImoKbKfIa4J6ZCEQt4HdpnTykOIwHvHGHuMbNAEzLG7DSYJkduriL9Vof9Q0waNPPSKfvtwMWTa7zNjLtC1FU/wSi4UmomjaNuCW3AOEspkRIzXdMxjnuGXWbZndK1A96Ad46xB6pnGIT54oxtv2E/7MDAyfKINMsE54hjh5GB9dWeH4xXLJeZVz/YcHpWaF3FyQ4YmS09P/XBlxjj89hwgvEzxPZYKeRyyGPRUUi+3bZXSilT7oW+v01oVYIk/7jfq7VMMzyo1Ux/H0TE8O4STW+0FHW6FGtwz5/xiT/4HBfnT3njS1+h3fTYkumq6lIHKrlU4iby9PUL2peE5cmSeHXNt7/4Jbr5krsf+CBiDYOoiiGlyL7vp9+n6KjHe1rn6Wrl4be/w3f++svY1Zomwc3lmvO3Lkm7iAWcMQSgyapUGJx9JufOjz5+DDM9fZOch+CF5dIyXzp8Y8EO6iN16MzO6iYy5kgqYHFIhZoKqZ8OhtYrGgdDrgmRigtBw3KcJftM6FTK0eZAGgVrhdUqQ6yMkz0sS2EcKkZQgklnKDFrkIlMF5+dogO9UoKrkYm0exB0GkW5J+0fBSH3kbSNSOxpJGAXjtxbrETFzYufCB2RphXE6OpfH5MBPCtVGDP9zgJYkFoZY4+Ifs2ILk1SSpPsRTDWM+wqu3Vkt41TVezwTgOxdWSjUFP1xFa81yWR3M6TVJ4g5rAUKZMdSbWGQdQ2lksl5kwuI9BgpdWbm1OPaikG31S6pSEliCmSyp5+LNhekJKI44p+zCzOOl583x3u3b1DvxG++uUf8OTtATc0tA1IKaSY8S7gjCGXSCmFEAL99ppZ65n5yvr6nNX6krbp8M7Sl0jJBh8W3Kw3bPsd1gut9dPhkQiNMF8s8XbJsF+ThhXrsuPmJvGe0tFIxAExCTEbXnzhp3nhxc/SdWdEm5GaESrOGop1U2WmNkIdA2vUY62w2WzVjofB+aBLpvruQXZ4lFIB+yOUE4DDFvVw/9Kv3258Be584GV+9Y9+j4unT7l57QfMhoSfQsprzmienWGz6nnzjXNeqob5MWzfeYtv/MfP88uzlvkL96kuk8WRM+z7YZqMCD44rMm0NXL1+ht84y8/T//kEWYYub7a8Nab5/T7iKkWS8YXaK3FCCRv2IWfgEOva3UJYG0lNNDODE1rsE1R3JEx06C0quL9kNyEpVZHjJk6CrHXq8KKIqdlio2sNSsap8okA9A3pjqhZhUR56wZCGL0AKxVlw77XVSJQoW68DhxiClkMr4xmODxs4IJhTp5SAEo02i6VKUW14rHEEtAhohJkbS/oRanObJBL2zjCmLVv5mSMA5Z8UxOqykmeEHKKmoOnVeluxQN/E719lCkqu5QjBCmdjtKxLbCPHhCG9itM/ttIu8hO4tPEBowfqouRVswzbOok2yC29DzUioOIYnixY1ou2wnpFRFD7OUIaaeWpLmh1i1+1WxmGCpc9EUvApDXzg56si9sFn1jENms7nCtSvuvtjQnax55YP3ufvSB/jqF5/yzf/8iGFTOZoHvGvw3iGmEuOAkBnHSNd5juYtuQwKkqgJsPRDJkYhZ4etkVwzzlfiOLIu6gCyBoIzGDsypGHaXDcMKXF9OZJ6g0lQBxiyx4Yz3vPShzi9+4DRRqUb16qzLeuoTskk1lpVJNeo+tRcSVlx8NvtBmssrRiqE72BMQmXD+t1JvJLYUK4T+6JetgOozPUqUpXVl7FBccHPvZz/Is//D3+/H/+XxnfeowbIlIFV/XAmxJeWK83vP1O5YHcYyaFy9e/z7e++AU++pufoTk7YVuUPFNjArTTMbnSSGX3+BHf+ssvsHr7dWQYWN1sePjwnN1uj8Ei1WAw2i1MbpHsHLOXfgLS0LzndsbkA2ASWYpSOQxUU7FWKJIVz47BGkcaC2noqVnIfaWO6IFQEjYbxCWsB99qW5tLngKLdYGhJ2LSdHZTaDqLsajyP2pWbQhMGOty+MxP1VjFWA3rriajmphJ2wJKAC6ZGicSSxUgINli0sg8QAOMY9bfQ9dipFTYiUbviQNxhZmzasAPYChQlZQrRshZD6TDMFsvet0gm8mWVqY8VDuJQ4utYBJ+ZjlqPL41bFaJ3W7EDZYULT4ILrxbwQpT9WpgWiGiGsSKE4OtdjK019vWNwT9vS0VqRqKnojkfEMpe5yZI1Kw4qFNWDHsQsBgyEPDft0zriqubVgcGXwXkARXl2te/UDDgweGT5/c5fiu5T//9evs1hsWbo5YoRBxjZmI1AkRyzDsSTWRYkLw9OvMbj+w76f5bdU3dxw2mnfiNG/YOUcxic1uo9VrLkhuyanipSEPsN+O2FooznFyfMby5IzBQCYiVccLRQlbSJo8xEXpM7VaPexEZ3AiwjgUdkYVCUaCemBNnSxdei1WOeQtqcrhwBcsmk57ey4ellKgwv8slma+4J//+q+xe/uCv/r3/yd1vIKs4FsVxOs/yMB6vUYeVx7Yu7RSePidv+O7iwUf/hefpOlatjVPKDDBWmicoV6vee1LX+bx978Pfc92veHhoydsNjvUR12xVRcptxAO52hOT/nk7/7OP+2B8194PPtDL6gsommtVnhet6Up66wPU7EVnFM/bimGGDN5rJQIcUj6fhUti2vJ2FTwjSVHxdvYoBnspSRy1oONorogsbr0ULBhgQmbbVVHo8uHrNWJCyCmYAy4TnAN2tZVxSVVzbrTz32W6aDU4XophaHuSHakaSB4g/dCDJV+r8sUqmG/zeQ0cBw9QoP3KKbcastZ0OwLYxWMmqNq7rxXA3jKkZqSLlMwUw6qOZRnqoPDUHJBvNA6rVi3N8JulUjbgu8tvgEfCq4pmCATjVpvAO9uEOFWNjFVocY7mNr7WgrOTuoyh1Z9qZLzQEo6Ow2uwxvPqjf0N4aH/zDw5O09sXfkERoP906XSJuJ9Hjr6ZolYgcWdwo/96sLwvz9/NVffJfr656QLU0babtALRBTYhcLeXpexQSctDhp6NpEKnsyMFsEdv2OZmZo26ChRlKxvui1FzLjPhFoyVmTvUxtaENgNmvwR4XazJkft5TmmCIemTa09XZOitJZqlBzJVbR/Fwg54wR1ZbmVOhJmFCx3hLMEqonTml5OrqY5nnoss1ON0Iy5FoooLNWM12HpYLoz4pGaE9P+PTv/g7XP3yHv/38X0LR0G6L0WUf+r1zqlxf3mCN4Y44Wtny91//BmG54H2/9FEcwjhdEQ6DHUe+89Wv8fprr5G2O/brnicPz7m5XiHFagZSzTjRv0WqbnrbGT/3q7/Kp/71bz+Tc+dHH8/80DOmUiWScsJXR7B+2hAKOReFO6KpU2WseshNJBSqkJNMPK/pT0WDSIpWJ1Uyxgu+NYhVyUiOqjBz2rdhrZBGfQ5V1FlBtlNClwb6uKmNxZZpnqbPX6TcRj7qTXfKxSh2+jlajeVUtAK0IDOPiaJsI5fBQuyhJsPQV3brQr/t2a4dz/UNR3dhlpXd50pRRX1NmEYPUxGrPl5JWFGyMKUSS1LWnj209u+KWKvV18c48BZOuobFkWV3k9ivM8PeUrKfhMkaXamob8E4i7PaZuk8Sm8EysTPVKsfLqoKxa3oh95awVshZ0OUwjDcUGQkx5Zx41lf7Lh+smXYGprgwVvaJvPCCy333hPwC5AwEJxjv1HSyWoPu7rnwXvvwhBYzDxh1rNardmuhJIcJUVqq0l71Ql9HBnHQhwgCYiFvgwQDD4E5kcLrM8sjh0vv3KH5XHg6ulT3vzhOe98f0XfJ5xJ5Ki/n7OFxo8UvyGWPalssSZSSjNV4NrS1ul9I6tGstSRKjtVyydBwzDi5LiIDNsVwV/g7T2cu0+uLcVYqkQEpowMfZTJl2uMmeAWddro11tSyiGIKRshesfilQd8+o9+n4t3XufR334LFwuS861/+wDgiKVycbFF5Jq7otGe3/nK/8fs7Iy7738/eQKJNmJ487t/x99/9avE1Yphvef80RVXVztK0sPF1DyxK8GiEqoUDC///Af51B99ju7ln4D2VgfbTMTgQhx1SC8G8jRfMMkSk2G/Hcm9hmfrul63TcYYtXKhF0LMFZv04CxFqKLOgnbeIFIpKXHYdVEKJU/fR6bs7yRTQI9VKm5r8LM65V+gXlvr3iXBTvYgi5KR06QZEFehCLVavGmRmqFG3S5TpoWBTFtQKFFAdDmTc2V1mdhtBp7bBJ5/sSEtR5qlwbUZ22asUQdHBaiWGDWrw06i0DrNcqQehNM/quNX2UEFcOC86gS7hWd/A1dPIn2fsaOQohDGgm0rtjFYReHhnG4aa42AQiK1By6Kfdd3WLeIE8jSou+d8RoIHceBYRe5uagczwIf/tCczcqx34ykkgkLODrb8PwLZ0gYWO8uuTh/wmoDQwrcufcqv/LLH+bR/Q0PX39EEzKuFeZnlpQW1Fh5ev6Yo1PBNe30/BZcnlf+/rXH5CwE0+IMnBwd4YzD2EzoIj/ziw/46Y/cx/vMsLnHc6dPuTn/Kh7DUddhzZ7NZs+9KlgyjRko8SF19wbSvUytAU0my6RSSClRYiElS4yRmLaMaUvaV0zy1LqjyAZTI+KhsGKbV+T9jNnsfXSzDyPcJYuQJ/nQuzTid6Urt7O/adqgguY6WSHVrZSssG8t9z7603zyj3+fP/kfnxDfeYgveQrn1lm0CIxUxjjw9PwpIol79g77p4Vvf/E/8rH5gvDcXQS4+IfX+dZffoHh8py423P+8JzLpzeMOWGk4qg4NHDKknGmkrxj8cqL/Pq/+0POPvJe+lZY/pOfOv/48ey9t6IBIRbRoJxSJluOnSAklU2c8k3T5CUVIdZ827KZSZBMnZwdRUhRYYZKMKls95nNptdqwykh11kLJU3tSlU0famAavnGlLSdTGBMg20caYrXM2aKd7TvTvdVha5i65iiagerwRRHiQ1x3ClVxUZyKbrt9ZpdoMsPrUKzKFrKCeRkOH8nsl0V7tz3nNwTuiOD6SslR8ISXKPVraarGaoUzPTaSK3TXG9CeaNShlzy1PYCaFxlteBcZW4dvvFsrjKb68h+J6Te0caqJI02Yb2CDXRbrQsMXd4IJR90ZVptiDl8EKfZPZk06cLEGNpgOF0mTlrDC/cdZIc1S/YpMrCjPeqZzTZKxImZkvfMO8+Ra5k11zSm5cG9lrLJ3Fw/oY6Vk0XL8k7H1fbzT6FmAAAgAElEQVSa7k5gPjekOmBD4PRswcV55bJ/TBwyH/ypB9w5u8ud5+5w9/QlgvOM9SGL0y1hfoORRNuc8sKrR7zy/mOuHl+yaFTds9sP5DTHpUKXK2W85vqtb7C89wpuPieVQskjKSWGNBJHFUfHtNU545hJ/UhNTzD1nFYuwayotUfY0vqC9Ibzq2/RzX7A8w9+DWefB1lQxCrMourmHvR1V+AFHKxh+pgOMXSEUwtEb0jLhg985lP84utv8Nf//v+gXN1gS8bUonpIMXgURZVS4vLpFaEVFsx49P0tr32x4Wc//ZvkFPnbL/4Z5298F9MnVucbLi+uybniBJ39IrhasUUPwRGDHB/zS5/7LC//ykfZz/0kmXq2jx8DRFTlyGV6s+JQtHoAxrESk0bEKVNfK7+k49nJiiOQNFQHoN4STvQuqBwKo9jrXg+USMFZaIJqy3IGRWmXqT3VGYvq3SrOO7pFQJx+D6YsCqODCZ0VFXWHIDJdaxVxOuvKJRNjYhg1p9RZg+BITNotrwuBgoCrFKtInqYRQq3EQbi5SmzWkeUTy90XHMf3PBqWnHVDGLSy01Bmff2M1aF5nS52qYJMkXwaV6WvmXVaoZUKRQyuNbTO0HSe+bLj4mFkc9lTxkoTBRcLpa2URhcU1sr0vQ6LDJleT72AS0EPYJhat6qDe3QmKuIoccTnyOLI4ULEehBvGIolMlLzJVTPovM0zZIhjYjpKXXHbnVF2i2ocUcjPU4sR7NMa1cwG7ixW8JsoHOBnEbabsfzLwV++TOnbDYbTpYr7p113D07ZdFtsdLQ95FN/ya7ckWMI3FoKNLy3Hs3+KOegMNEz75ktgMcVU/JA227ps8/YL/6JvPmHmNZKm153DHmPcOQyMNAjj25H6kpkcenUN+hkbfw9oq2WdH4Ae8HZj7QzM5obcP15qtcPobT5z5OsO8lMp8kJoWS1b+tYVQFa+y7Sw45ODX0j5SEzYquytbiz0745d/7LI9/+Cbf+8u/wuwHTIqUqvP0ULSyTNUwDPDw7SvuS6U7iXz3b75GNYacEq9/528p+x3ry4HHD1ekaFSLVwVb6xSkpa1wcYEy6/jwpz7Fx37nt+F4qR1USs/s7Dk8nr0Nbazo+1MxRlPRx0HdGRqiY1V0bHRWFqvGJNbCrUSCKU1d0M2rDtwMsb5b0osYpDpyUVZeNZDGiLVTtsS02hfRD2nJh9bB0HaWEIQyJPI+MztqtEm0eQr4lsmzq3qx/CNiXkRb2VgGXc5gqFUTpayL1KTzsDyJTJkYfGmoVCs0XnBB9W9jb7m5qGxWPc/ddNx7KXBcIKcRv4jgsoYYGf0ZOWedS06yhzoFvkidGGmTBqUmpqrZKia8VkJQ+9QsBMJswfpo5PrxDf1mwEehyZaawDSZ6nXRU6vONw+vfy3TGairId2kWtHTuKizIkU9IDEw7vdsriK2MeAKdTp4XYBSRuLYg7RY45gZjczMOTHUjHOWEm+QXLAu4LKjv9pSSZzMWrIUXPDglEgsfuQ97+0YU2QYzkms2cUdTdfTzM4Y+yvGvGK9ecp2v9enaCun783cfa+lMyeM1x3f/MojHl+1LE8Huk6JLcuwIdtzZLggRseYE33csk83xLHHx0ztN0i/Reo1jkd09g26cE7T7Ji1lc4ZjImEkPGmVwnWbMfN9dd45+aGl179l/j5+yl0ZGPIUlRhV/VGbzGad3yowOv0XxOdW3LGjRm8CpfPfupVfv2P/oAnbz5k9d0f0kXB1kRvkroHq52EMI5+V3n8cM99F3AMvPb1r2m1uYuMa3j8sGfYO7wRPBE/dROWQpMFsZ6da7j3oX/GZ/6bP+To5RfYG8GmgouFZ/149uLkw+FWK7WoYFMlGHJbNYio9/PQjQmCFa2y1GStq3+dbemcTL+fkBLTG68l/SHF/aDgFEm3w18xRv9tmQLFqZMIVMGj/WbL2BeOjhydcdS81pDkCmL1znpYGDhnp97PTIE0lWEP+62CDIwtak1ykwhYucc0TaHtFOpp8KqEKRnrB4xN2hrFzPk7WzY3nvvbjpMHnlmymK4nzApSI7kACG7K1hBTpoSqSX5TdataCkgRxClKK5dCMgnJBkE0gS0UbONplyesn67Zr3qGbcGOBjs6kk+YoAuOMmn0FKOlFaR6Qw0+eFLJ5GTw0iLVstvuKePAfOYITcewG+n3SdtCI5Rk8SFjQ8bWSqkDqYyE1kHOOCpiLWNOnBwFSifMugX9GLk4HzDes5Q5jbXkQZifPIeIpesSYZEZk0PWGcPA4ihj3IrdPjKMWza7xM06UW2hbQs2RLqZEJxgSiQNR1zfWC6eZl592RBToh/Amz05nWPkISk5hmLo055x3MDQU/oen55ieQfrnxDsOW1zg3VbvKvMGoc1mhNiMNQcCXaHaQpmvuX8Ysf5OwsevLLE+RegtHoj0buXqgWqht4ba1W7yTTuQG2KyU7jh6KLpxiElz/2ET7xX/0uX/if/hfq209pskVI7E0hCzgKVSq5ena7xJNHa+4+OMOK3t3iuvDk4Yr9PuuBZyKuJkQKjkqXLb5aRjHMX7jHJ//rz/H8Rz7A6HUDbUolyX/xmPgnfTzzQy/lCbQ5fUgPA/c6zSp+lDRz+CAKyn+TKpRcifldTboAztZpuzUp3afhvVLIJiIE3GJ+DosIJt0bVW4dBzYIszbgjWefgFIZx4F+GJmViYV3aB9ExaHGTgLPgm7tsiGOldUmMuwEays+HBYB2g6nWlRaUwoiGTdJWqhWZ1y+4n0k7gr9TlTwO4y8/b3E6spx730tyzszXMmYZsC4RDXKh3PeIsFhTEEmpKk1lpx0iw0HEjMIhVwSYxxVFhEFqYlsCv5EOG4Ni13L7jJyczNQo8E5zUyQkBUE4SYHwSRaFjuNDIqhZostgTxYtjc940bJwcboouiom1Mu1sRtpmYoPao8S2l6rQr9LjNuM8vjgHHKDjQBlgtDHQyLuWez7ThPsNlmHrx8j2JXPLm8osQ5L7z8HuYLuNq9QRKhbU8QCmPa04Y9+36kVGGIBesD8xNHqVukOhrTEYKnlsD9F1/hva/O2Dx+nbwCvEGMp7hMCRf06SH9GNj3nlx7XLyE/SN8vmTWXuDDE1q/wtgNNujrbaqf5rEqryoZbN0Da0qMSLHMu8Tl5Xfwy1c5vX+CMWFqYydNXtVRR5qiKcVaKHmiM0+V37TV1R9QSE6QheWjv/1pHv/gh3z7//pz5HrLvBpMzexVpq4Z0aYi1bC7GTmvK07vn1FL4fLRDcM64bF4yTS1aoVYhRmGFkM0Qj5Z8gu/9Wt86FMfZ2ih2oxE3S7fWoyf4ePZAwdGnT/k/K6xWoy2aNYJNWclG08JS1a0zE65ElHmfhHHfjr52uBIebL+TLSUCreUEFDUu7H6M3OpE15dDx9r9bAs2WAc2LbQtI6YRnLO2FuzPoib0sumjk04VIxaaQogWZDkGbdCHS1S7eSY0K1uzrrNtFbb6ZQL7hCWbDK5jqRakdZhVXFAawx51EI4psL2euSd1zJ3X55hX+holw7cDkKiiMp+SgTrDca6qZXV+SV1Gh3kRD5EtclUGWIZ4w6KxzlHNRGZRYzLdM5B23D1NLLd65bXeg0t0jQz1RJaLwTvkeLJvSf2sFuP7G52pDGqVU+EJJE427GYtcxOA/OjjjQKq9WOsbc476hFCKFjFoTV9Z7VIIjLhJnSYhocm23mrbcviOMJpj/BSY9zie4MhgxXl0+4cI62u8vMz8jOMqQ9Ke/YDiv6YUcbTuiWd3hp/iKRM2K9JMYGkzytmeH9jFl3xDw94P2vFH549feYXYW5o7SCzSMxR7bxhjE+VuH8cI6v7zDzTwndE7r5hs4nvC0UO3GSq1eXgk2AY7dDM3fznjwOOh/OLVUCp/ceMDs+1khNiYC93V1MGnmdf+eME8EaN6XyTV3OpF6oFExmukkZmudO+cTvf5an33uDq698G99HQhWiFOL0vYOpmGlhFW8GrllRSyWvIyGiDiJBqcCAF4uvk/6v87zw8z/Dxz/7W4SzY4oxSNaOSrukZw/Ue/YzvcGQJtz0IfXcGd0uqsxIxZWloLYxa1UOVmFMiThJIcZpFjVGg9SCEyE45ZiJiIpxa50G+zrsVbmLm+Z5ZkpzL9QsGqNIpuk8YWYZc2Q3REIrOOvxTcE2aRKIqsXncOCp59doC14hj4UaDV1j6JScTsqZkixinZq067u0Da0ArcpZcOTUEyUhrdB4p6AEVyCDz1r1lZh58vqW3fXI3ZfmHJ2dgNtRbY+0UJp3lyvWGawpiBTETHimCcZw8NMe2qScE7VmpASMVZ1ZMRmaRHviuRMC60vh8ulA3CbNAnHTjQRo2krbGFIvrK/2jNuRPMRprghUrUiSFNJxxYdI2xm8CE32ZBcYc8WIZ7ftGeIIJdBvK5urTNN47Ak0C4dkS9oMrB8XHr/9hG1sePlDZ8wDnBxFjhct1yeGq+sLHv1wxd2Xz2gbj5jEkBXYOvQD+7RC7AKRY9LQYuxd5o3QLo9YLJ9jNpuxe3rN5Vtvkm++z/1lj89V4ae1ksuMnEfq/nuUeE1NO1p5xLK7Yd5uMXaNsYmmaSd5ltGFXinqwQZK8rRORzxljNSxUkog1jO64w8yf/ALSPeAKjNU+3C7t5hmbwfZil771pppXn0I45hEB6je0hQNwMIZHnzgffzKZ/8Vn3/rEfmdp9RhINgJfKojdJV+GYUm7G82ehhmnekaFEXvq+rx/KRdza2le/Een/r9z3Lv1fdQrGCLLhcrU2buj+Hx7Gd6yVBSwlg1YRsLZSwotXhyERZ7ACGTKuQiDCkTs4b6YCLtXA3im02EotVLiIXGO7W7VC3NdYulmRGpVCBBNRPxQiipIjgwGWkyy9NjXDCsVj25VNquxYZCNcMkEdFKrZAnkbK6EahqCh8HBReAMJvXadEixKgb61QSMdrJ+aEZA2aqFOOE3DZGsCWTBUyj4TnFV5XSZKEphhQLMVU2VyO7deb4TsfJnQ7XOmw34BcRN6/TrNKCt9gDQVmYfs67Zu9aq85ZkMnyFKf2SCVGpWaMjfiucHJvRmgX3Fxltts4jSCElDLbVUFmnpuLgfXFnpn3zPycNIzkFCcxuMEhkzZXsEtDNYloK25pkWSp2UE/KPkmiWrN+oKMlX2FtE8420EyGCJNU+gWheeWQtw9oZnoN8wK9IXX33jK7vqG0xeWMBfaE4/4RNt01NhSRo+XY+xoscHSeE9jZuxWax5//+9YvfmQenHD+Oiak2BwVaUgmYopI66s8XUEeQc3jzRhReP2OKsOm1orsY5Y7DSoVnDDYbFe8gjVUKLSf3IORE6I7gHiX2TMS2zyGKfXDodBxbTBlXoQ7DOBRaOiwQ4z68NIp0KuSa1uAoaEm1k+8slf4u1vfpNv/d9/gWTwFGwRhoTyAY0CN2pmckNpcLsYsEUIxeArBAdGCsULHDX84m9+kp/5xEehE7LJ5CnJr6DA4B/H49lLVqa7QBG1N5nKdHjAWHSKQBFyFmKuDDmhWdw6m2g6z+mdjpM7M8TA9fme64sd+21kKJW+KMKmMVr5lemHaGU2VWllcnAAtZgpT6LSBcNi2QCFoS80rqXtBOMGvA14o/NFo+NAcskooelg9ToYycEGw/xI53w5qRxHlfqGfjiEEzFBFQ7C4knwe1iOGMHYgptbaicahp6gjlXbywjjqMubi4c9F4975seB557v6EZHSCOmTeQmUYujOOX6OQPVodY8A34KCp+0PFqVFnV2MEl6KGZq4TNVtoRZ4E6Y0W0qfd+TosEkQxqEcYy4YOmWDi9QEvT7ihQdIQgKJxWBNGTSrmJnjm5+xMIGxlG4vtwSugaLIedI20ErgTxAzSNjr4SdxVHAN5V777HEVLHhKUEqedvp9jgWWhHuzgKbTeTye5fsEZ575Zj5fXCdI/aGzZMtR+2IMcdcPe5JaU1K3+P0dMWsjtyhcnkTqauCPQ6QdXFjquBrpeYVs8aTSYS2YF3Um68YZDLcY3TsYopolVVBqn7wU0nUJKTRkaInsSD7u2R/j5QDcRdZ2khjEhzmsSKHPnVyUxw4e/V2OXhrJdS9BmaSLaWSycZgDeAq/sEJn/iD3+Hi9R/y+Fvfoe6iZkGjM9QygWuN85Anf7YVpEacQEANR6YK1YDMPa/+4j/jlz73W3DWEW0lmSk0q1TN0eAgaH+2j2ff3t4ebHmavel8A9HDcFKkMMbCmDJjrVRnCK1nsWy58/ySo+cCYQEItMuW+fGcp4/XrNca3kwqjAhNqQRvOTiyZVKda1KYPp+K6HNy0LTqZ92uR4adsHj+/yfvTYNtTc/yvOt5h29Yw95nn7G7T8991N1CSAg0AwJBQBCwcBGIMWVTJo7Jj0yMwbYQ0pEg4CIVg2WBQRMgYkygkvgPlaLAJmAQElgCtXqS1OpB3X3mYU9r+L53yo/nXfuI/I+oau+qU6ruVu/TZ621n+8Z7vu6G2wzknIm5YQxXgtVSirFOPKmQY6FFIUQhBA1s1Ra8AI+gw0BwZKyYAdDCJpcm0KpB5hMjEISofEW56TSTnSZn03GTy0lJkqj0AVTAaExGEI0DOvM7uWBxc3A/Ljj+B1T+mMRM1lRpgnXAkYtYzZVfaAUfKN/DoNKeQR0dkqoDU0lrlr4YiLnhLFr9am2Qj8Kw6qwWkYGU6CMuKbBdw1pSKx2A0OJSCg02WGz7kPxBcQTkxCHAnaB9SMpWoUktAIxsk5Lul6x+QZ7tMF3ruC9+lvFBoYhMY56XV9eD4ivO9wBZk2DTw1D9Kz3ViwuHLC9vY33gA3sXrvBjcUB034L12XaZmTe3eB0n/BDy95CWN5YaNGwieIFaz2dN0iJjOOKYle0E/COoweGtVVHWoScgyoQ6hoHAyVlYhZitsTYENOMWOZkf5zodogyo0RDWK9wfok1WzRNy+YtunUH+OKjQD76e0dQjFLUG1wlYE5qvIEIuVrDdl7xAK/5zm/lj69cZf3iDQ1kLxExmSAFEY0j3Y+aRLhjqhSLhDcJL0avx94yueMMr3rbNzE7d5bo9Pco6OHPIkhWBe5/Ejs9REjZVLtYvYCKVVFxKeRUiCESs+rKXOtpJo7t4zOOn5wy2XY0c8F2+u1aB+I6bOdpry/YvX7IsAwM68gYEjZlvNc3Q6SGCOVbqnaDVTCog/m8xxrH/t5A2zqs16elmPqElIhIQiEFRq1fRcfXnAsxGMYxK9bHmGrDKhijFGjQ9CjXGUQaYsoY4ykR1suR/b3AsM7EpC6SXJJermVz8Y5gFQApFooX8lBAdB+aU8HgGVeZq88Frl0MnL6r5+Sdc0pYEycjrq2Yo2xJRo8YOWu3bRThUguLvr5WrPIEUZCBEe12i2Qwyi/sekvTGvqJZRwM+7uDRgU2Bteq/k+cYdzLxGWC0dKIo8uWWdMjTWYxHmB8IJU1i8NEiR4rGmBkG/39FecfKwSikGUJPmKajJHIzCtrJgtVr6mvv+vA20BOkZQMx2e6I7WLzHxrDk2LvXdCXlma7gDjlzSm4FhS1pHFVXj+6RWLw5GtmYFJxs4MvhcKkRSS8heKdrONdToOluphzgVrG9IGSiGVli2ZkApjbohlSnY7iDuFlW3G3BOkJ+WeMkLOa1bugMbPMNbiXIcRf1T1hE141eZAqOMzR2DSVF1B9Y3Nal/cuJyKFWTiefBr3sCzj3yWz//uH+LiAus2r6OjZMPuOHI5DhgEWxpON0793yYrV9HA2Hfc++av5oE3fBW0lmJ1V7nRi1Igy2YU/09AnJzyqDThoqrvJCpDqXcJtVcVNat3k4bZsY7Zds9k1tBuW9zMIK2h+M0StKgIksKcDtdaDveWHO6tWS0CY8ysc65FT1t1K4ItuouwkrCSaH2hbQzLw0NKHPGdIQ4KDejaBt8UvayZGv5TfY+bQwmIPtEVloZkbfN912KI2lGQSX4jGi00xVDQwGTrDV3fszgIrJZJQaDU/V7VExqjha+YAk2pxGf18qYa60jMCGqFW+0GLg5LVgcNp+/pmBwXymRFbjO5KTqeWHVSeN8ohisNuvyWgm+dZovkpNzCoisG530tQhu7W0FM0pS6JmGdZb1OjEMgxUgRsK3QTGB5NRP2CjIYnINwECFGYk5YcRSbGVeFHAreaqeJMWSbtMMuKnJOqPPFE7Gu0QjQHJVgbZU2kkMmRTA1bGo293RdQsQiHkxZk26MJBwuOCbzjpAOCGMm4YhjwsQtfHMbwTyPTAN2nmm2wXVU50gFqlqLsSoTFtsgJqomrr5mIYx1Et0Up8wYC4EOaU/hu9tx9ixjmLFcG5V0ZYdkj0aQRsbVAaGb07Q9KXmM8fUhpJVvc7iQSmSpCxyQXJFkmmMrRoXMUgok3f8VK0SB5sQOr/2Wb+Tao0+yeOJz2Ay2OEyx7IfIlXFgT3SXbcPIBGGrcVgUHZVc4eSD9/K6t34j7tg20RRKjpRUbZtGYyKzUONJv8QFiL+RQ0ZmjFafbgaSTSQSKep10hrNUuinjm5m6GcN3bzBdgbTGWgspvE0vR4yxpUGeNsO3JDo8KTSkk3Bes9qEVksR1LYBN0IFqtJYgkaSbQ+Mmk9bWsYVwu8FCaNYzrz9H2m6wEzVOWSYIwjogWLUkixaORjEHJQkEAKCd86bNGOMMZ4VKSShu3qZTqrts02inyfGUvTCvt7IzmqpzgLOOeQYivxtuK4RMkrgsZemlQwGVx0rJaZ1TIp3TcM7O+NnLqz4eTZKd1WonSZYnU8laaCG4KQi+r7UgqEISot2tSUrvrfXyqSnLQRlm8yTBTKObHQTwshGIYVrNeagpfr+znuWvYvjSyHiNm1uHWGxjDkqOLi/hh7y5GYCu1EU91mOxMO9w9Y7C2R6n5JuRACsAoqOM96sDJZC3rK6qlWxbpe112VHJWAXl8XmbGsKMZhfAOpZbEYiKPgSsfJndugO87sjkMW15ZMjxn6maWYgVhUBmSIOG9oe0vTWYqpJG3ZEHiK2sTq7k3lRy2JnnZ6B667h1LOEMIxFd2bAUckplEJQ2LIKTDmBev2kK49jm31wIStNB0Vu+rxwmzcRhs4wa0hOFcgbbFGQadFVQ7kQjKO5Au3vepBXvVtb+Gjly5Qri6wWG6GyIVx5CaFoXZpN3LGjwMT6Zl7T5GEnNri9d/5Vo4/fC9rrwJ+m5OeXUTUwlayRh+UcpT78aX8+pIXvZ2dCRcuBdbBMZCJEihWuV6d87SNpW0NTWdpe68WJZ9xvcd1Ht96mtZi9TOEb2y9qoKLiZATphXa1FBi0CKUDOsxsY6BNGi2rcsaUWhs5sypCXednePLyDgEslWjfYqF1XrAR8OkskFTLsrOq3vCUq+2JQlpzPprKMQR0pAoY8S1Rj+cdtNxVt5dyUdCaYUuFDIjtsB0S4XKKVhFIo2JYa0CK2vNZiWk/7632C7hciElkOQZ08hqQHNO0d3lhadHlgeeM3dO8NOI6xPtFMpEu6giETYJb0Z5aSGrwNsYOfqzguCs5oAItgYI6UgllkpRVppO0xqaNayXhZSE6axh2Mp0x1puvDCyu7vArYSmh2ZqoFP46LBa1UByTzPptAsMUIIQQsI5cK1lWCdC0OByQWGgsRTSmIk5QypI0DFCKTdOj1hFHSljzIwZkoUYRiWMlBbXzmln26xtxzis6beEViz9NCGdfmZxerBqJlIjCQBTCOMAJSr9pqhCgE0+MoCZ0XZ3c+LY/bjuLoa0xRg61VdWCYspDTlZyGu9xFqHJIirpHnHHsRBESGhRGtjbgl+pf7eIUSc26yRFKuWayFOplCsjrcmCTmqsDxvdTz0zV/HZx/7LM/90Sc5uL7gUhjZJWvB8y0UWIWBGyVzaRyZ2J7psWO87GvfyL1vfj1p6kkUiIUihmx1yiLnI3mTsV/6fR78DRS9O84e49K1qwyHa0YB02YaJ3hr6MWpEBJlxFmvua7a4QFeBcKZhKtyi0zUkBon4AXTGXx05CBYmxGJtBbEWbAaMBOGwjiOlKIZqLefPcZ0CuvDFSUlKIYYICepBJWIMQ2VGlc/WJrAhaDSmGx0ZxwNq91EHFVwnFYJ1yZcA671xOoocMbo6bqOcAW0o7KKdGpsJadkvYgOK1gdBshWu79Rw4S8q09OVZFiJw0hNyzCwFA861UkqUWdNlt2LyeWhwc0E9g+0XLsdMtku+AnI/gR6zXZnuo2yUV3abZ+QKVKapDNU1qqGwatwjU7uALMMTbTdjoCxyCUPJIkgDOcalt2L0b2rw6ElZCwtAWWiwXrdaTpGpydUGJm/+Y+acy4KtehFM0bdkYT7FCKThwrhh0NlbdFpwuyaIobmXHIVQ+aSSUTRRjJ5NUaP53Rz7dwbU8MmeVylzQMuLLH9CR0faZpM84peds1Wuys24AYOLowlKTSKVv0YDWWBmlOMjv2ELPjX4Fxt0OaQrSIRMQcatdmRgKRUlokG4VTRIPBkQdYLVaIP6SbzinGA0Kqk+xmd5ax6Mlg0+XVR58odiqnrJBTK2RxYAWTIlISuRimp+/gld/ybfzFJz/Hi5dvskcmGIObTHnTN30TYRz5j3/0R4TlkhtkYhy5b+c4r/qWb6PdOUXIWX9eREf/YgqSdWdfSCrGl+rI+hJ//Q2QkwvHT7YchgVeLNIYvAOfLQ26bzPWYJ3DeleLlVWvqCjfzogjRAUrFvEkEtmAeKsWF2fVXeGt4qRItAIiluA18Lg4FUrunOxp2sLqcMF6oelaFMO4SqzW0PUGKQ0lCyUFCCq3MTU5LWVROxECwTLuG/aurijR4J0ltQXXZGxbcI0FZ7E+I63uyKw3GK9yhpBDNboL0gqUCFn1hn39/cI6w6CFWwN71LaRkxCzIRrPKsNoHesSCEGNeLkmaHUUciysF5nFfmT/ZuDUHRO2T7XYHmw3YlzCeRTudeYAACAASURBVKngy1JlRrXo2c0COqtroxRiTDrCiRbuUqocqdJfnBOM1SyNGCJ6RE6IWbNtHH7ScHA9sF4l4sKQjSWj1/uDwzWtN5gcKSFB7RJKLsQxU/QZRYqZUPeepQgxGazxDHFNChtnTyHnSIggOIxtKBLr9bxoVKgEUl7g1iNJIIxrGlkzm0e2jkEzha5vFYYZB3RLRgU+gDEZa7Ts51hft2IJqcVN72F+8pX0269A/J3E1OnKTQKkJa5YvHRE58Cta96GkEPSz4EIJQTiuCKPS+j8EdChiCEWsGaDlKq0HdGdrT5YNRhcd8Smdnw1p7kCQCRlcjQslolnr97g83v73MyRwRjiZMI3vu1t/NA//SesVkv++U//z/zZ7/8BN8aRwxRIu7s8d3WXO1aV9uwKxW60jFK98GBNpVpnjYj/Un99yYteDgPOJaYzS0K7FC8OES1q1juaxtc4Rqs9PIZUDEUcKYMlkWMNzDRGn+Cysd3o99EHjKXxDTSRoQRs1qelcQHjVfTrfSKPA1EyJeq/s9ESppAx0iLWk0mal4AKNZFSfbOVG5cSZew5vAk3LhVsEaadEJuEb3U3NdqkGHgrzOcO7y3juqhg2BaKy9hWcJ1KUQrqPmlqCLrYQtNZwlqw60IYM2HQPVZKhlwctp0grcPNgcEQY+FgiMS8CTSHxhV81kCi/TExHB6ye82yc9ozP9HiJoJMEkayCmvFVGaevr5GqtUv5TraWi2q2Vat2KhOjqLtXxZ9XcVTw9HrQclknAs0naGbORY3Iaw8Q0g4aZhtb+GcZVwtyWPCl4IVo6HVRjMgNtrIlDOQcM7gG8vWsW1a33Fz7yoJ3ee2XatrkQIhwBhhvXZkEUoZSTETV2tF26+VzOPMyHyncHwH+nmknXtdV0jGRKUGqeNHdZtJMlYyHjDZagaLmdLM7uHY7a/Hz15BkNsoZaKaz7LWSFHZJ40vUEbByxbJTggmEs0ByS4RGShF9DM8OMaVp+m8RqCahlIsYHSMrcVtYzvc6GE31stSoxFAd7LmyCWi1rHVGPnDP/wTfvXDH+HC1ev6fSYTvulvfwc/+va3c++5c5Rc+PF3voufE+Gjv/8HjMPA8sIlfukDHyZOHW/6+tfivf4siejEUEo9ZlQMWSnVRfIl/rLnz5//kv6GT/6fP3O+ZBhSQkzCbeITrS6DTGNxjcM1FtM6bOPwbYvzvn6goaSoeRc1Pd6YipEPhhINKaCXu0HzLuJRwnzBkPA2412h88KJqadzQBo3Vl1iAdMYbK+yBD/Lum+SqMvazRUu61hnEMpgWB84rjxfuPLCqE4DEf3z5YIVh7f+aJFsskGS5nqEdSEMOoJpdKPBN/rDa52h8R5rBeNq8phVoKnzBufq9U6EMRVCLkRrMN1EozBFGENiGBMp655H6gdRiiDZEUZYHkT2b46EQWh8xwYUqtACJeNQnO72oHbQAJZhZRjXDis9YYyay5HTEbFGjKEYU61RqlNTMo3QNBaxmbYrTOeG7eMTTCsUW+jnHdOtOeAY11G7uxpoI1bAWVX4C1inD8uud0y3GnZ2tjEeYlkjDTSTln5rSiSQTSFQSEXAeIzzZNQho5j/zRY0cHwbTp8S+lmiP+aYbPX4RsOIiiQNjsJgTc2noEARjQENnli2MLN7mJ96FdI/zMgdYObYIrg8YuKSvHyemxc/xtVLf87i4FmEFd46UuwIyel6QqpkyOoeT8TinNMxu2rxTP18ayTpRrdcjkbcjSNjA90oR92WksOlWMKQ+MM/+BM+8Isf5sXPPodJBrqer/nWt/Kj7/gJHnjwQcQ5xFpOnDrFgy9/OZ9/5mkuvXABxsi169f53HPPctvZ2zl75x1Heztdk6Qj66mCj/SzOJ+efff//5Xn1teXvOg9/jvnz3eTCcYmUgpHgdLGmDrC6phqasi1dRZMpmk0u1bTxqCUSCZTkoViKSmTxsi4juShkNaZOCZCVKCniGAlY50WvNZnjm95duY9eVziTVEPZBbWY2ZIhWZiabYS3Tzj+4Bp9dKqPyB6tzPFQCwMy8Jqt+XiM5mDG5HWWaa9xZSkfLL6Q982Xhe6RS1q4zqRI2iauRaFtnM0Na3N1GwLqubLVNmM8QXfSg050qxaxfkVxBvwQtM3tF2HbZQrOKwjMVWJTcWaQ6HEohfn0XC4G1keRPJoMclXwXImhARFKuHDHGnpUxTi6NnfizjToEfKsRYATzYaOsMRC7EWbQMUzUwQo7s/cQnsSD83dHOh2AFx+Wg0sx5sY5HW0sx65ie2aec9ueLojXWIK7jeEhlYhkPGPFC8B9eTxDJiidKQcUeIs1wiFJWYGDH6YLCJM6ca7rjDceyUpZ0bTGOOIKq2hv8YDBaNL8g5EmvXnWJDlgmpO43t7yH5u1iXUyATvCSassTFA1Y3nuDKc7/H7vVPUMIFZH2TvLpOzgcKbig9OTpKdqoNpeZhYHCuoWlbDR+q3u1SNrs7HWmpaqovjpLUCaLozxai3toshCHy0T/6GB/+lx/ixSeexgbBtD1v/vZv5Uff9Q5e9vKHVUVA/bbGsHPqJA88+CDPPv0Ml557ARMiN69f5/kvPM/tZ2/ntttvq1bQgnVOcWSVeg6ACFsv9aL3yL9+93ljPcYVzXi1HlCicS56nneNx3mHN04V9z5VuYfOJpmg4yDqjsgRTUpbj+QQkQHSUMeeWEghIiXjrQaMTzvLvIdZ57FZGJcDnVfowDgKB4cJ4wxbJxuaWWSynXGTiOt0L6tULIs1FpsgjpnVfmH3kufFzw+EZWF71tE6owtqBdpRJKvGzujTOupxkWFI5KzRmNYVpvNWdYIlEHOoqHftkHLFahhfkNr5ta1Ga3YNzCbQN9C1AkYJKW3X0fZ9RYALnZ+zXm3ArfnIpWKLI47CepE5vBk53E3EwUHxKJetHPk+CyjAAPXwtq2n7SzWR8QExKjaP5LUkACoiHsj15GjgsfGw+kMzhfERsRF2i7RTXQ1YFyp/s+sxadriNaxCoEkEeN6ViMk62i35izzyDJGsuuw3SmyOQ52QnETxHSArkNyXFPSSApRr6dFydJnzjScPeuZbidsn6p7JENO+oDxerhKISoNCF3Sa6ayIzOhuBnFbrGIPcvgSSniS6TNC3y4wd7FR7j2wr8jjZ+jkT1aEq6ofjAMN4hhgfOQy6g3rxovmovup61ttPD5BuebOrpyJFPJOVX/bb2s18AmRAXHFBWfSxbGdeCjf/QxfuXn389zj38eE0Galtd/8zfyI+d/kpe94svw1v019JsBxFiOnznNfQ++jKee+hyXX7yAiYlrV6/x3Bde4K577+LM7adxziLGqM60lKNCKAJb0ztf2kXv8d/8qfPWFD2jZx0ncjJVmKyLVyMWZxxNZxCf9LBhbVWxB4o4hmGLEDqUQzZSQqIEIQ2ZFCMkCKtCHFT8a3LEG/AGGpu0MDgDYyKPEcFSsmU1wGqdca1l67ij24r02+B7lBYsyuCzRvn/ZIhrw3pvyo0XJlx4esQUw4kTrRYe0e8lHj3bO7VQJbTgOedxxuKd1U7NFvxEiCUQNtmkVhfPG3KGGHVl3Bp5teg1HXhf8DbibcCagDEB5yLdpKWddFivwuHlIjOOSqoGywY7XmKqhxFLGAwHu4n1oWClwxrNFNZ82VJBlRkIOJcQO4AMdQ2gdGgVQKMUZ1HU1YbwhkH32DXExhpFfW3C4L0TvFfbl28N7cTiWyExMsRAMZ5kelyzhZgJdrLF5MTtuNlpTH8SOzlNP7+bpruTpj2FsVOQGkEQAzkO5Dyojzbp4WU6azhze8ep09BN1tguIY3Svg16US+iZJ6S1dplLBjjIHticIyhR2QHmDHGhuUYWK8PyOtd7HgDH3bZv/YkN699HMkv0MghLYJHtY5FNCAhh5FhONRO3gpS/bS5bJLPDMZ5XNNhnUNbOi16m8uobLhTtcBscFMG3euZYkgh87H/8Od88L0f5NnHnsJEQZqO1/1n38CPnX8nD73yFXjrsNWOqJGTikZTNqLh9G2neeDBczz11FNcvnARCYlrV67y3Bee56577ua2287U8bruHKkXeIGt6V0v7aL32G++53wpiX7iwBZChHGsfD0KMerYZH1BHDjX4IzD2kKxkcSEG9cmfPLPd3n28wsm/RaNh5LWMAp5tKSgT64wjITlSBoDzhQkgxWhazOz3tBZy7AYoIJHUzasR0Wv2yaxddIw2S50c4NvDcYKxmjH6EVR3Wm0rA6E1V7PhWcKNy6PbM0dk0mklLGOooJpBPHgW6dPvFLw3uHF0HmP9/aoI3StPTJ4C0Z9ihXHo19yFHTkjoql3nyMB+cLbQdtk2lsom2gaxOTqVq6qOLj1SqwHiBlS0YzM3TNDDVAgxwt4xL2bw6sl4USLd60QI2WrMVLbDzqapWVVonVdb+kha7ubmunmDbLbSNHYe8iBWcVh6WjZMHYhDEJ7wtNB11v8a2laVvmWzu0/Qna2RkmO7fj53dAewbf6S/jTiLM0WD4NaUMpLQmpRVxXBDHgRyVvr19vOfMHT1bxyK+XdP0Gd9VNJdFQ7zrLtQUZRLqqGdJ2RNGTwgzMCcoZcp6bVgs1gyrPeLqKnFxkby8TlpdYFg/ibXXaNxKu7vsMLboe1OnvyJAGihxCQgxecZYSGjhSxjEt/iuw/tO8Wg1RmEjBRHZDKP6ZcQcKRQElf/8x49+gg+/70M89anPaMHzLa/5hrfwo+/6SV7x6lfROH9U8EIp7O/tMo4Drm31SovKrk7ddob7zj3AZz77Oa5dvIyExNXLV3jhxQvcec9dnDp96ijE6Ei8TWZrevdLu+h97nf+2fmcM65x+M6xGgPDoDyxUlB3ghM9VGSHs4p9SsBqmPLsM/DHf/gCf/Wpyzz77B6XX1zg7ZTetypGHiJpLYyLRBwCUlPnlWzhFIjZJbbnHRIgrBKNbUhJO7x1dKzGgdkOzE8W+m1DP3UYj/puq6xGsiFFDRJfLQr71+HpJ1cMq8LxnZbW6/XTNQbjlfRrPDStU6JKSgiF1lmc0Seg6ssytrEU6kUyZUpSXZmCSDlySJiKparuHi0aTpFKrjVYl2l8pm0L7STTTyKTyYp+mpkfc0xmZUO3rxQY3XNJdS04Y+ouUfTPeZhY7EbGNZg68m6w+4hlA+bNSY4OIZAVWlp0+S7YI2P8Jk/DbGQWde1kbD20SDn6X1eT24xNWJtpW9FQbgLioJtt4bodpJmDnYGpFI9QKGMgxTUxHJDiPikdEIdDUhhIY6BthFOne07e1tLPAk0/0vZgnOgYa4BKBZJc34ekbhyNKHCMg2Mc5uR8hlJ6VqvIsBoIwyHD4jppuUvLkt6vaPwBxh7im4B32m0hiSQJsU5VC0aJ2kjWA0tO4LZIbJNoiEXI4jHGYYzH+RZrvX420f30JjAK6shb1JIHureMIfHJj/8lH/gXH+Czf/UEZgTjO17zlq/nx86/ky9/zVfinWfzTqaSuXrxIh967/t49C//ipc9/CD9pK+edmVf3nb2LPc98ABPPfUU1y5eghC5dvkqFy9e4s577+bk6ZPVJlexV6WwPXupF73f/mfnS0Vzd9MJ1hvWq0gJemZHBCcNUtzRD/GQHdf3PJ9+/IA/+/iLvPDigiFojOPu3sCFCwes1oJ3nf4kxUBaRRhL1WRGTM44U2ibzPHjHa1tWeyNmGzwxpIiDIOhaRusCxw73jA9ZnGTyGTu68UuVRJM/dBEIQyG1YFl97Ll8gsRbxq25iAErC00reAaUBNjRlwNyqYcHSC8t0cJapnKs6sXWS12G/OGqXs49Hpc8tGHR/+WHghU6SNquXNC02TaPtNOEpOZYee45fhJy4nTcNc9PbedndLPlEA9jCOFrOQSKRhUgqHXNkeMsDqMHOwGhmVGisOIPSrAVECC/jdXiUIt1oDmONSnfU75SGco6M7Ke1e1dhtqiH4PYwzOCt5pVrG1Ge8Czo0Yt6KYBUPcJeU9cr5JzgdI3sPEXSQvKemAFG6Swx5pvUccFuQwMmnh5ImG+bbQThOuuh30Qq7i82KUAFeydiiqwc5KEg6GMHjGMCGnHVJuGMbCarViXO8T1rtIWjNvDTvbnq6N+Fb5f9aqcE3f70yxOv5rtrCqEjCGgCfIMYo7TSzHiMVXlJQe0qxptOg5h3NOC5+pQI+qWdHoIF1lUAwpZv7y43/F+3/hV/jsJx9DhgKu5ZVf8yZ++B1v51Wvew3euUrt03iDyxcu8oFfeC+//cEP89gn/xKM8PCXfzlt3+nnEJW+nLnjds7edRdPPfUU1y9chjFy+dJlLly8xN333supM6frA1Bfy/nsJb7T+8xv/fR56wzDkElZ6CY9MWlGQ04G0qbjSyQHh7HhCxfhE49c48mnbnKwH1Q0iz59pGRW68Tlq0uuXhugNBpxmNeUOCpyOQqGjGsi85lnPpmx2h8ZF4G+0cZ9cRCJKWNNZmvbMplnJltKR+ln2q1hFINl6hiRE4SVZTjouXnZsnd9ZD4VujZhSDSN4NqC700teJvLZcF7wbpyVABTLqyHSIgKO40l147OYEy93G2QTEWtRCnpMSKlVJPI6kVZVCQrRrDW4hpoW3UPtJ2l7QzTmTCZBY6fypy+M3L2ngmnb5sx2bK0k4JrSh3rbIVS1iB2o9KGFGC9zCwPEsNKwQzOeJUP1ahLqELUWvSMeMQ4Da5JSYWx6L6ypI1ZXrBW5aP6g2+P/JnW2jq25dr5qbDX2hFnDrFmDys3MFzH5qu05TrkQ1K4Qh6vkYYbpOGAuDqANDJthe25uixcm7A14lKp2zW0vNJaSs0qrmTV6lXWghdSR2GLYTSs1pH1emAcFpS0wJQ1085ybKul6wJNF/FNrEFRVMR6DUS3eoG3Vt83MS1j6RnLDsWfJZRTjKkllYgBbQ5wWOvxTUvTtqpvrc9Kpaxox2fFIbiqdBA+/YlP84F/8ct8phY8sQ2v+uo38eM/dZ5Xv+F1eO9VioWmC169dIkP/MK/5P/41V9n2NtlXCx5/NFHScDDX/6Ko8K36fhuv/MsZ++6i8cffYzdK9cgJC5fusSLFy9y/7lznDh1WgulMcynd7y0i94jH3nPeZFMCIX1OOIaoWk860UkBfUIJpsZnOXm0PG55wOPfOY6l6+tCOMm5EdIVmkNpoBQSDGzfxi4en3FGIwSQmyAEjClYFGq64mdnpIihzcWSoi1mTCo0LfxlsnU0U+FyRb0c+imQjtxSCvVSM4tHVS2xJVjtee5/PyaYRnZnjdYG5X11otizZuC641m55qMd0LbGZzXriwXGMdEzErXGGuaGtyy6UgN9slJL6abkPK8uYTVUbFQw7czYFTrp5o+h1in8oamYJuIbxP9PDOZZSbzyPZxOH7aM98RuqlCEJw3laoC4xCIVcemDxz1Ba8XmeV+Yr3M5OSqBcphxLE5GFLUnWCtr8TgVAOrFSNvjUUqz4+jfVRtdutfbwgi1pjaRaoUyBnwxtBIxMmaxq3p7ZpGllizxMohzh5gWZHGNSlEOm+Y9Fo4vReMz1DRTA6DNwbr1OOcEuSw6ZioHmRHGD1jcIy5YRiFYSgMQ2AcluR4gJc1WzNhe25pu5G2S/iuYJ0CETaRppuO3SlDi1IysRgOx4Z1Ok0zezlZ7iSmiQIUappewWCNjrV60GiqDIX6wNOHphRl+pWsBe+RT36K9//8L/LYxz+FGQrGtXzZG17Pj//0u/mqN74e9/8peDeuXeNX3/ev+O0P/SrDzZuUuuaJqzVPPv4Evm158MteTtv1iKhu1TrLHXfdyd333c9jn36U/Ws3KGPk4oWLXLx8iftfdo4Tp04Bhfns9pd20fv0r7/nPFjGsTAOkRQLnZ1q2tKwItjCQjpeuOl4/JkFTz9/yGIZtTIgZDHQdzz46ldz+q67uH7jOiVtYg4LQ8hc3xvYX0TENnSdwcuIL5Gd6YzOdhzeVOaejgaWmFQ02XWOlEe6iTDdciqVaQp+ZhAPqQRiCpSiAcZpgOVeYbkHNy9nTHb0rQXWekntDTQZ0wmut1gvOjr5+sEX/WMVHMOoMElEx8VSDCmmqkHQQqB5vVrQDKLjx0ZXVrtC2SDC67I45USsKXHGNIr2kagA0EaF0NZq+JFrEr6JtL2i17eOWdquYIymnjWdChbWQ6pdGEe4crJlWBb2qsCZ5DA0mNwQQ1LHStGCfdTl1S5XFRabPBPtYEGqVap2jNZURb8ePXJRSo0eP6yG7OSmjnaJQgRJGEl6re+EtoWmMUx7R9cUjSywejFXKZJ+xpx1aqmzFms8TnwdZ3WlEoLUfFvHEB3rAdYr3e+GcU2JS/o2cHzbMu2h7SL9BJp+895/kW6u7i2PtHRZWI/CwcqzPx6jnb8K1z7MmOaEuHltLEY8RSwFixiH8Z6maWhq4TticxZ1zFgckoXHP/U47/+FX+LTf/ZJzFAQ63n4ta/hR86/k9d+7Zs0Ta2enrTgXeeD7/1F/s2vfID1zRsUkzDzFmmsWuKWtfB1HQ992ctpuo5NfIB1ljvvvZs7776HJx5/gr2r1yEkLrx4gStXr3D/uQc4cfI48/lLvej9b+fPpyRY44kDDAslfNjGEKx2d09fhMe/sODKzUCO2gkUDMla5mdO893f/w/4J+95D9/xXd+F7zqeee45louFJhCiJJCDVeTmfkCyY943bE8tEyus99Ysbo4YLF3XKJhTEm2nHYhxmWYi+B6yRIpPNL3RYB6TsJWvaYslLGG9MFx5Ubh6cc28n9A4i0jCdxZc1nhApxYydVAYRHIV8SoWXwOwG01kM05lLNXxkDc6uopzMkavv7aOfJs0ezE6epb6/80bmGTS/AQpGy2kLsiNVTnJ5nqqO7RUMzQE3wp9X+inhm5q1ErXw9Yxz2TmtfspWa/BpWCxlGzIQaqvNzAuEjmAKxqWnnM6unZu7Eemym8y2vlJUpF24Zb3F6qNrXZGGqJebu2uTCKLaj5L1QdWxj3O6u5089feC12jdGPrDcZp8dFGWTU0xSjUNsWEE0+r6eOMQySMhhAcOXdg5mB6xgHW60QYR9K4oPUjO1uGSZtpu0zXC01ncI06aaRaUozVEdrYjZOjMEbD4bohcJaTt38tfvIwY95SvHtWFuBmL1eKRapN01iL9Q3e+/rg0yIqWTBZd3ifeeyz/PLP/xKf+tNPIIOGpN//Fa/iR9/zLt749W/GN03d4ennZ+/6DX7jVz7Ab/7yB1jduE6RTH9yzrf/l2/jvgfv4wtfeJ64HhgPlzzx6GP0kykPvvxhfNsc7fiMNZy9+25OnjnDE489wcH1m5QQeOH5F7l67Sr33n8f99z3ipd20XvkIz91Xkq1biUd6wKF/dFyeb/l8WfXfO7iioNVRlLtVsRgZj1f/sY38INvfzvf94/+EbedvZNjx4/zVa9/PfedO8fla9e4cvWKyg+qfGyIhb39wLiE1nSQ1WdbQqb1lsaZI49i2ngQfaGbOdreU1ym6Qz9pME1WQXB6A84yZIGz+HNnhefWZPWmWnf4JtM21tFDvVOqb8SoWrQnNOFfK6B2EYcMRuWq8R6nUlBSBW4aI1ajTQ4SJf+WVEhFFQfpQNt1UzJF11T4RacNesO0Igy+2SDLJdb0ojNtbiQq2m+YF3GNbrf66cN1gE20E0L821PP7c0ra0I8i/+Zg4pnjAoSHRcw7DOGtZUC6xeflVsnbKmzEk2EAVJooVvM9qKymlyLrWz46hoVqMKiapvkwp4rf2Kqa/75lpiKjoLVCVgavHbyGw2B6KcMyFG4hgpMRHGxGqViVHH2hh7cmoZ1ooRG1YrSloxaSPH5jCdFCZT7Y6VwqJOmoI6JYytTiSxWKPX8piE/ZVnFc9w4o4308+/gqFskU0mpUhKWZP8QqIUUyEKGjJvjKXxLU3b4Xw9LGEUZpsyT3z6Sd7/3vfzV3/6F5RVwriWc1/1lfzYu9/FV3/DW/BN89c6vN1rN/jXH/gQv/G+X2J57RrFZGZntvkvvu+7+fv/9d/jVV/5KlbDwDNPP0MaRsLBkicfe5xuOuXcQw/RtO1R4bPOcte993Lq9tt44okn2bt+E0IddS9d5Lu+9/tf2kXv0V/76fNSP3BRLPvBcmPlefYyPPqFFS/sJYbIUYpXMpbJqRN85/d9Hz/8E2/njW/+OrrJ9Gjh45uW+172Ml77xjfg2pbPf/5p1os1pgi2CCnDwSqxtxeJo8VYj1jDZGJpfa58L6OL+pJxDubbne7xbAGTaHujo4nTBDQrQo4wrj03rziuvrCmbxrmW0I3DbhOjeemsYhXMas4qiAza7dHvW4aS8YTimWxLIQkLJeZ9Xokhog1lrZpq4o+HRm3N3uulG95RdXeo50eG11fvfSq51Gvq+XIH4H+/2vB23SVkI+Q7GKzFklrQaxGZkqgncB0bphveSZTh1jN+UiVGSemkoQlEkIirAvjWhjXegUtFGIAyZ4cqV5VIY+lHjXKURpbqiOvulFQJ4HVi27Omn9cku4HxdRVQL1uSqVOi5EjW5ZxolnG3mAbJcCou7Fgqh/YVM1gzoUQCnEU4mjJpSXlhuUyszgIhCFrwYsr+i6zs+2ZTApdqxh9Y4uOhA5KRbNv3ptSdPGv0kXLOnbsrXY4cfvXsXXi9YzlGJFEKpGUEjHEI9cFR79EIw9oEHG0vlFPrrUYDCVkPvP4U3zwfR/mE3/8cVgnxGqH90PvfAdf843foAWqXmlzKezduMm/+eCH+Y33/SsOr1whS2Jyeovv+N7v5Hu//3s4fnKH2daMcw+eY7Fc8eyzzxLXI8PBIU8+9jiz7W0eeOhBLaS18Dnvueu++zh12xkef+xxDm7sUsaRF55/gR98+0++tIvepz707vNFhNF4DnLPhQPHky+OPH115OawEWYKSSylbXn4tV/Ff/9P/zF//wd+gLN336PK8/rBhin0TwAAIABJREFUB/RDbC07J07y6te+hnseuJ9LV65w+coVzf+sH45VKNw8TCwGoel7uonHmgwxKWgSwXtL2+mxoaBABOMyriu0EwGXqgNCiMFyuN9x5UVY7K7Znlv6yYjvA64vtNMGXKm2s6K2ubqYb6yOaRsjeDYNk+2T2H6bYnsihjHCEBLjGCkZ9VjaqsWr9UtT3aq/oZQjTd3Gn6y1r+76kCMCx+a108JpvggCWtQmUuUmG+lJJtfsdSElw7BONI2l641axaa6r3KdIYsova9Ug7wxatsSAEcYhGGVWS8i66XCFogWgiMNENa5auGqyDkLkm2V8KBXVLglxRDRUTTdIgXr9Vqv7N41eiBBMfJIzRip0h7n1b/c9v4I4NC2Fu8F72o3VZzu8ZIhYghR2N9bK9Q1J0grJl3i2JZhPnO0TaJplbNXRN0LGKnrhfowypYUC2OIVXM3IXA7W6feyPbJNxE5rUzAUqVMlVKdgsqmctb9qK4tPCIKAWhaT9s1qm/Nhs9/9ml+9Zd/g7/4fz4Gq0gRz32vfCU/9JM/wde/9Ztpu66OtLonPbi5x299+Nf4yPt+kf1Ll8mS6E7M+La/8x18zz/4Ho6f3qlqZJjMJtx/7n72DxY899xzxPXAeu+Qxx99jOnWnAcefBlN11JxBjjvufu++zl+6iSPPfoYBzd2kZj4H9/5rpd20fvkh376fHBzdmPPM1cjjzy34Pn9xCJVLy2GYCzNiR2+/e/+Hf6n97yLN73lLfSzGUWMaqRKYXl4QBgGmqapWifBdx0PPPQQr3vTG0lSePqZZ1mt13UnWFgX2Ftn9pcBax1tO8UZDzHiTKJtDJNO8zAyA74TXGcwfcbPBGmy5txhWI+Oq1d6Pv+ZEWcsp056PQR00E4apCn6hC6bEB3N2PBWsAhWwJtCEYt0W7THbsPNT+JnOzTTLZpJTzHCOkVWw0gaI9aqJ1mLX6kmc452eqUGPedUu6Sa3aHdQTkqZHob+SK+mbUYY7Fiji6qm/W3FEtJqt7PJRNDIazVK2yN4LyGoPuuMJl5uolFXKBQYzY3HmUHxhYVxxZHjoUSjI7AqwzREQcgq55MicPpaH51NZRIAQHcCuIpio0v9dDFreMvxgjeqh5MR+nEhlLtmo0URi+RYhUE6jZHJtSlQRaMOFK2hAzDqKHtcVQltmFkNikcmxv6Xj3PzmWc13WIcfW90WFbLWQZStIQpcNFYjF2DOUk/bFXM995HYkzFGlJOR1JZXIq5JjrBT9VDzOIeEqpT0MDrnE0TYe1Lc89/SIf+sVf4+P//qOUpRa8u7/s5fzwO9/BW77lrXR/reDB8nDB73zkI/z6e9/H7oULJCLt8Sl/6+9+J9/7D/8eJ8+cgAqY1atzYba1xb3n7md//5BnnnmGPATWBws+95nPsrWzw7mHHz46jmjH57jn/vvZOXmCxx59jMOb+/zgO9/x0i56/+GDP3/+yqLl8ecXPPr8gqurxFggIgQxxMZx9ytezo+86x38w//uv+WOe+7BuFvYvxwjTz/5BP/8Z36G3//d3+Xue+9l+/hOtTfVru/kSV73pjdxzwP387mnP8/V6zeIWbkgmcJyzFzfH1mvCs5NsNbgqx/Xi9D4moPgIrYDP7PYadE8Cqd2teVyi+ef3eYTf37AmRPHmM8y3gb6ieCmgmnqIYGEc4JzomJla2icRjzGUlgmg52fpj1+J6U7hp1sY9qepp/QTGc0fU8xsBpH1suxMuyqJc5SEfRSjxu2op82Oy1z1LWVWiwKOjpurrtHlzarBxQ2u7NN2FFWVFeJUAKkoHSYFDSEHaN7Mddm2olgmsxs5plvOSXp5KgjeNFlvXO1u6pdlbV6zV2vE6tVZLVWOZPqyjRjQaTgvcXVwlyJVdhqpdrs9cqmoNR/nqKSdhQBptfXTYd8NO5CzdfQ1YXz4J3UPZsQRk3WG9aGGB3DKhJDJoaESGQ+gWlf6PuiD7zW4hpT1xFQxVK1SAslQgqGnFoODgx7hx3N9AF2Tr+B+YmvAHsHKU+qNlK73c1DLEbNlig5k7Pc+lWonm6wvsH6nhefv8KH3/chPvbv/pSyiiCOM/ffx4+9+118w3/+rfRdr0Ud5RIeHuzzOx/5Dd73sz/H/qWLZCL+WM+3fvfb+L7/5r/i1JkTKtUy2ppAqROEYWt7m3MPPsSNmzd59tlnKGNgsbvPo48+zqnbb+e+c+cUCMytUfe+l51j58QJHvnUI/zAD/4PL+2i95s/+7PnH33+gKdvrDmImYQQMURr6I/v8E1/+zv4x+9+F1//zd/MZDYDqaNXzuzfuMH//X/9W37uJ9/Fn/ze7/GZRx7hzz/6Z3jvOXvXnXR9f+RQaLuOcw89xOu++k2swsgLL15gtVppEaAQUmF3GdnbHwDHdN4x3wJrBiX1WoNrBTdPdMcK7dQgjVVgY+hZ7c958WnLcnfk7rOeeXdA3wdsW3MinOiHv9y63FoDVnRPlsgsIxxmT3vibroTd1HcnOJ6TNPSNA3OtXTTGe1kiu9bkjWaB7wOxCpnEUvN17gl4NWIyM2iX3d9uh/T8HH9B9SlPdrsQN0bJQ1tyXr5LUU0MSspqTgFGAeVboAh5iqXcYL4RDuxdBPVO25tO9oOtVhVaKRx+tpKLdhNa7BeL6wFAbGqWQyFnIwSoqNg6j4xhqxZGVGPYWTdAUoxVA/7LVlPdkjSK6yOhOWoAG5wJCXV9DijKwArphYVhbOSIY3CapkoybM4yJRokVKY9Zb5zNBNhLazNbulauiKYxgsw9ozrFqGtZAHiINjtXTs7xvWYZsTt7+Oux/4FiZbr6ZwhlI6wNaHU771K2l8aE6pdvOwoQ6rizHjXIMxLZcuXOe3f/23+di//1NYRcR4bn/ZOX74ne/grW/7W/Rdr/GP1IK3v8+//a3/nV/+X/5X9i9dohD/X/LeNMyyrKzz/a1x7zPGkJGRQ+ScWQNVFFhAFRRSYqM2iIJKc21QLEAFUeyyFfq2NmInk9gq4gUfm8Z2uNB6FWyuTCIgIFNBQTEUUPOQWTlnxjycYe+9hv6w1oks7pf75T58qBvPc57MjIo6cU7E3u963/c/YWfaPPsFP8ZLX/UL7Ny9MwNIuctLGrfLR6ZQdPt9jl55BSurq5w6dZpQ1Qw3Nvn2nd9mx655Dh4+kvh/+f5U2nDwyGFmd87xhOuvf2wXvbe+9nXHl8aecYw4BE5IYmE5ePVVvPo/vIZf+ve3cuiKY6nrIK3vXNPw0L338q4/ejt//vZ3cOHBh9GNQ/nI6qVFvvKlL3Hh/Hn27d/PzI6ZtM8gdX07du3kxpuexp69ezl56jSrq6t58Z+WtkMXWd2q8V7T6XYxRUSICnDYQtCZUnSmJaJQSCUxAtyWZ7ziWDm7Rd8GdnTHFGqMbUV0KbKFd9a15pjEtIsTE7YBLkqGXuOKKWb3X4md3oUXLQIKocEaidHZQLVso9pdil4f1eoQpMKPA8040GSDBqHE9m4uARSJ5iOl2tZdTjo+v43qpmKWbqiAbxIYkWiB2S3ZJ4fo9P9rYrDU44h3aTEfZULnlM18t4z6KuNpdRSttsCWqbghyUaSSV+rtEzqGZFAE63A2szJQ1DXjiqHLLmGZA7rRXKGiZnv5xOlSUaZLf3TdRbDxKpd5HD3ySLz0f0t24g3JIJ09CErGRNXUGWEPOmKFa4xDDYF62uOmak2ZaGSiUR2l64qz3jkGQwiWxuCpUsNmxue0aZjfcWzuQ6N6+LFPLsP3sieQz+AMFfg2JlJ3QnVnlB2nGsIIWXWetcQfbLAilFkClBAy4nMzHLp4iofet9H+epnbycOGhCavVdcwa2v+01+9AU/Qdkqc4eXwr6Hgy0+9L73864/eBsrp06l636qw4/85I/y8lf9ArsWdiFEArYiPqPjfnJmpPssyyCnpqc4duUxFhcXOXPmLNQNW6vr3H3XXezcvYsDRw4/qvCBtoZjV1+FNfaxXfT++HeOH2/I46xUtGam+eHnP4/Xvv63+eEf/zHaU/0UuAMQIptra3ziwx/hbW94E5/76MfwG5uozLmCiIrgRhX3fucuvnbHHZRli737FijLMq14hKBst7n62mu5/slPoaprTp85y3g8Tp0SyW14ZbNhabUmxJKy20aXjrLt6PYk5ZRCtQIRh6+hWpOsnanZPF/RLxTtLGEyHYkpNcoKYo7j2c4nEFlBolKH4qJgHCyt2QPM7ruaYKdw0hC1BC22fdASD02DbSHLLqpsY1stdKsLGOo6MKp9sk4njdQTHWx6jmTjk0xMVd7zMbnliRnhnoxQ2Xko/d1l23wfISpi0HinUwFyqVBoLVDGZ0uozPfL9BGp0sGRVC7Q6qQu1LYMIUYa57c5a8ikE0528smtWamEsiLANRHnoHYJTAkho8AOiBLhMydNyERLkeT1U74zibkjSje8+C59b9oFWmOSSD9zHJPzS+5KBYQgidFSVZLBlqPM9vPjkaOuIqOhYzwKNLViPARXWzbWHeMtj/AWVxsGI0s5fYhDV93Mjr03EvQBvJgmyNStIRIBfZIx7GNIKwLvCMFlz76I9w6CAx8QQaGCZeXCBv/4wU/xtS98jTCoEcKw++hRfuk//AbPfcFP0m63t7W0PgaGW1t85O8/wH/9/T9g8eRJIg4z1eKHnvcj3PLKW9h3cIGULRpATJyP8/WzbRgQsuFE+plPT01z8OhBlldWOHP6LDEXvgceeID5PXs4cOQwyqh84Aik0SghH9tF7x1veMPxBgjGcODqq/nFX/s1funX/j3HrnkcyqZkJ0EkOsfD993Pu97xTv7i/3gnZ+6+D103yaxRBCbaaWJExYj0gaULl/jKl7/MxUsX2buwh9kds9muOgUN7dq7lxuedhPzu3Zx5uwZVpaXEgWEZBG/UUWW1jzjSjE9N8X0vKE17Sm7Am0bwFNvKUbLU1x4qMYNHb2OQrYclIGirSnaJlkETXZH5N1RtjvTImKUpHKCgSuY3nMVvfkjONkiSp13TeSFe3K2FVISlQZtkbagaHXQZQfT7mDKFkIZai8Yjhuq2m8bEqQCeDnSUSqJVnpbuQEib/+yG8ej6BCXC2Byk4lBIaLFO8Vo6BKXUEqkcInGo0TOwIUJvJ7GepIjigVtPbYUWJuCYUbjJoMs6euV0YljppJ2WJuIKSKmAGOTlUwI0LiQ5HAZ50j5wAJXhyztSleRlpOsW5UI2uQbTUqMzYHlIat5MiAkszUTmQak1LbZOjFKxlWgbkhduIrgNStLFaMBNJWEaAheU41INJdKIKJBqy7CzLLjwLVccf2z6O95IlHvo4k9giJpfEPcbkhTHQ74kOM5vU/jvQvJjcc1EAKhBuEMqxcHfPojn+PO2++kGaYOb3b/fl79W/87P/7CF9Dt9bYLXiAyHAz46Ac+wJ+89b+weOIkMTpUt+AHf+xZvPQVL0kFTyXj1jCxq+dyiNCEAhBFJMjLrxki01N9Dhzaz+LyCqdPnYGmYX1plXvuuYfdC3s4cPgQSut85QnEY73ovf1Nbzyuuz1ufs6z+a03vZlnP+/59GZnEosUgYiR0cYmn/n4J3jL63+HT33oI7iVNWzwqNzJtI1gYcYw29Y0LuScilT8xoMB3/nWt/naHXfQ7/XZt38/RVEyYbW2Ox2uue46vu+Gp7Cxtckjp0/nOMi02ho6wcV1x/K6Q5UlU7vb9GYTwdRXUK0b1s5Gls+MkhfeVKCYkuiWQrdlclfWMROAk3AcQ8pwlWBUMg4dOk0sdrBw5fXozhxeFnmnRUJQYwShSIFJKqs0CqQokLqNtAWy0JiWxba6FOUUSEvTSIYjR1W5LPIX26eyyBSSyfitsjnp5cAYgAmqmy2ggkxB5k7SNIKqCgwHGVDJeQyJMxMQ279lsY0cKykxOsvctECoAKJJ+RiFxRhNJNnRx0hOwpNIFROSqhJ3UpkcF2BkVqvETNZOrJGmjlS1wDkgCKJPyOtEpaOQaZTP4eTBi2xaO+ni0k5TCpEPhvROlEo3JyF1ummf6WnZiBYaVxdcPDeirjTtVgeipK4cg80aXyfn7igsRW8fhx7/Axy67pkU00dxYg4fu4nOlBPVtk02BYBP3RyR6CbZHYLaNTTNOK0cvED4gpULm3z6n77A12//JvWwJkjN3MFD/LvX/SbP/+n/jU63ly2nUsEbDUd8/IMf5k/e+nssnjgBOHSv4JnP/Ve84tUvY9+hhSTLE367g8v7ku1DYWIWMBlzk1gmjb5SwMz0FIeOHOD8hUucO3MOGsfmyir33H0vew/sZ//Bgyit/v9R9N7/t//X8V+49VZ+5bWv5di116IKw0SGiPecefgk737nO3nXH/0xJ+6+B9E06BgwRKwQ7Cgl1+3vcO1hy4GdhiljqUee2udxheRVt3jhIl/98pdZvHiJQwcPMT09s20VJIxmfu8ebnjaTczNz3Pi1COsrK1to5tNhKWtwMnzNVVTMrt3jk5fE8dQLRpWTtZUm46yKzFtiSkCphXQHZBFQBdJBSClpBEeaVOa/EQVUEfBZmNo7TzE3L4r8LqTeIki0zAmNhlCEtEgdbqyokaJEiktUUrQCmlLTNHF2ims6VMUbZQqssqjoc43S/Bhu4uYZGlAzBrNiJ4g5NlgUgYNLoUXpQIhGY18GuVyApgSYrtoENPYLrdLXwIf8KlTkhnlTaagEVtIylKl2IBsFxXjZYlZJCBVyO7J8lEdIEidniOh4mnXOunaohepSDeJFhJ82ukFFwjushmDn1haiQQ2xTz+JmkbGTlNiG/0MLETGw8bXOXQAoTQjIbJsCIG6PVa+FDjXINr0tgnjaU/f4TH3fBcdh+9CdHeS5DThNhKvwPhkxSQSe+UOirvHcE7ovfbI20KW0pEZV9HfC1ZubTFZz5+G1/7yp1UowYnU4f3y699DT/14hfR7na2C54nWV59/EMf5h1veSsXHngQ8KiO5RnPfiYv/+WfY/+hSYeXSPVMDjWStbycUJ/ILjQT38R88OW/IICZ2SkOHT3MhUuLnD93nlg71peWue/ee9l36BALB/YjtUI+1ove45/4xOPPfv7zmZqbI8qctRCh3hpy22f+hTe/7j/x8f/7HxgsL6O9RwMaQVsJDvQM1x1oc2RPZM+MYLYNHRvoWUuoAsMm0uRRSYbAaGvAd755J9/42tfZMbuDhYV9OSEMEIJOr8/jn/h9XH/DDVxaXubsuXM410BMUpxhFXnk7IjTZyrarSm6tmS0XLF+bowW0O5CuyMpWoqyLSk6AdMGXSRCchMcqIluFMAThGDkBMNKMbfvKto79uPIMYQyxYlPEEiZozG1tqQIdI1SRVJXiJiDcEqk6SClRukS25qhaPVod7oY26IJkfG4Zly5RPwWidKSLO0mBcZnuRekWEKILpGEU4GA2qWdmhCGymuaJhXnSec04QqSOYATVZqbdOKTFCzhE19Ppu8rdXLRtkZCfh1am21qTXJRTqoFISPaSooyRQcoE5NpQpbKTQKHZB6D6zpQj1NGsK8kvlYEpwlOpsV/loDB5Y4PwTZ6m+7fROcJPnW/RJ25cinDVwhotw3dnsJYhw81IGi8wPZ2svfok7jmhh+jt/t6nJnDyTKl0pF1zrCNvKfDwxOjS//du8wvzI4n3uGbJu1anWFtacinP/ZF7vjSNxmParzUTO3dy6te+xpe+JKfpdPrZ71yIoxXVcWnPvox3vGWt3Lu/gcgOmKpuPFZT+elr3oph48eIIoEVvh4Wfe8TVbPCpc4WYvkTN1J0E8MqVhfHn8jszt2sO/gQc6ePceF8+ehcawtrnDPvfdy4PBhFg7sx2jz2C56u/ftO66KIsXYAcJ5zp96hPf8t//GH73pjTz0nbsQdYWOAQ1YIZhrSY7NlxzbbZjvjOmImkJKhIvE8QirHN1OSWk1rgk0Lp82MYL3XDx7jtu+8Hm2tjY4cPAgvV7/MjvfaHYvLHDTM55Bd6rPyUdOsrm+DjmhPTSRxcUxD963yXhVEyuJb5LkaLrfotUG3XLYDphWSuOSmhzmM7lRIARPFIKqEYxri4sdFg5fhWj3cSIVvdo7GtckFDFkBDVTLXRMS0FHyglRKrngCmHzhQjGWkzZRrf7FJ0OZctiyiLZektL3cRkNx6gdv67FRvqckMpXIRMhK0aqFyBi20wbapgWdmoGVfNtrmnYNKqZyeYSO4sE30nGYOqbe0oGWxJ3nkydX3t5HKjlcJog89OzdEJEKmrM4XE2JQLIrIu2pYCY6EoJGWROkcpQebwHqmSZDDUMSlBhhFXK3wtEehsCgs5rIP0LmJ2uskUbany7yGhq8FFRgOfi3lECEfZjiAbhNY4WVDO7GH+4Pdx6Jpn0p67mkbO4KROCo2YyMXbBrDbh8YESfZMDoBIXj8EnzNMBLHRrFza4hMf/Ty3f/4OxoOKIDRTu/fwqtf8Bi+65efoTE1v8/BijNTjMZ/75D/zh8ffyOl77oXoEC3F9Tc/hZe84iUcOLxAjA7vGlzttyMxQyZ++3y4OR8SshzBOZ8Q5kyjcS51ocEFXFVnvbCn0+8yv3c3Z8+dZ/HiIqJxrF5a4r7772f/oUMcOXbFY7voeSGOQ9pbNcMhX/qXz/D2N7+ZD//t3zBYWsKENMoaoKskR+YKnniky9Hdmh1Fg/U1oSZFO1YOX6Wv73UkO2YKZvsFMkaqceKGiVz8Rlub3Pn1r3HXd77N9PQ0e/YuYKxNjhxC0On1eML113P1469lZX2d8+fPEZoGGQMywGDoOXVmyNJ5j1YFszsLpmYSHcNYh+0EsAEvwMfkfZckXmbbCdcHcM4wrjTl9F7mDl1FLNo0MgViTygIIk4uMJcudJ/CjXxoCKIG7bfF6lYENCH5v0mF1BplNNKmuD1jNYVtY4s+RWsaZTsEYamaiPPgckdJOh8Q3hKdwjvB2BsGTYsq7KDsHqA9dYCGFoNRoBrX4D1aqUzrmKRbTdbTZBuotJDPMUKJJkMai0QW2kMgxCZLtyTVuGE48Ay3Ir5JPnAAITikSooJqdK4bKzE6IjJOz8EyfdQC0yhMUXaI2qlIAhcE6jGyftvPPCMhx5fGaIzBK+ILsm3EkVFIoWBKGmaTHNykWYMGyuOeqyxpkycTK2ovWfoFHZqgWNP+EH2Xfl0TPcgDV08JgUKPcrtmrzDm3TLEzRZhMw5JEsMJ/ZaMb3OlYsDPvbBf+Hzn7yNeqsiCs3Unt384q238qKX3kJ/Ziapa0jPUY9HfPYTn+APj7+RU3fdDaFBtBTX3PgE/s1Lfoo9CzupqgHj0YBqVDEeVjTjmroaJ0PUekw9rnB1Q12NaVz6sxqNaaox49GQ8XjEeDiiHlWMBxWjwYjxaMRwuMVoPKTbbTO/ayfnLlxkeXEJ6QIrlxa56+67ueWVr/ieFj39//4l/99+CCA6z4WzZ/iHv/s73vvud7N46hTaOQrSeWukYMoqju5sc+X+DlPliI4eYwV4a3FeMHYVMUK3W6JkRNtAHYd0TMFse5qZYsiDlwasVp4mKxCawZCvfOrTPHjvfbz453+eF/7sz7InKz6EELQ6HX7gh36Eq665hr/+i7/g79/zXpZPnUY4hwyRzSpwz8WGxXXP0PUQbctCb4BtRSINMnd3yQ7OIPAQHVIpjNFZSpSI0fO79kLRphGJiCpCjRUJma7rmqZJS/mJ9xyk+EglAkUUKGGwoUFVA0Ll8V5SdGYRtk8jBR6BVB6hJcL2ca2GxgWapqIab6DqdYTforCesrREV1E3Y8YVUHnqylGLFmVvgdnZAxTdWZzQdNsjdtu9rJx7mNHyI4yGA4IRFDYZukLm3OlEvfC56/biMqqsSQTlGCcj6cTlxGNagf60SCho4xgNAsOtSEoaS5w+ISLaasAjtUsmoCp9PxkiRqfDJnWhElcofJVMUaWKVMOUwBeCwA8lmwNPpEGXoAowpcx+iAGlXRrnokcJj6uy8YBXmUaTi72CqDpM7TjIwhU3Mb//yQQ1RyM6BHQ60EJa9G8XPS6PttsmEmGCZttkk8Y4dcQUxCDYWlvjkx/9PF/859toNscgNN2du7jl1b/Cz/zCzzM1Pb2NYKeCN+bzn/o0b3vTWzj5ne+gfYOyiqOPv5LnPu9ZzM222Vy7mKzIokIJk7vOFDEaSSwCKRJxe1vcDBMt4PYoHHyyB4s+q2RiIOAIOKRU7N07zfN+8kf48Ljm9F0P45qKB7/xze9R5bn8ISbEzO/Vx3BrK97xpS/xl+96F1/6zGeo1tcxIebuLu3u9vQNh3e22DurmbI1Vo6RuOy7JpLl+EQKgyIGhySdmD4q6tBio7acX3OcuDTm/EbNlkup9g2CRghMv8eNN9/My171Km565jNptTvbeygibG1u8KXPfZb3vvvdfPUzn8EPR8gJGR1B3yoef6TNzc/ocd0ToN/bQNsKYSOOJo3vMl0sAg9B0tSK1S3HJn2ufOqPYHYeYoxOXnK+AR+pase48jQujYhNXVM1Y6LwGCNoFxJLoHBj6uXzjFYvQV0RgsL0Fth1xVOQU7upANesYkTKTHVVoHbgkfhQ490WIgyQGoqyRaEFWiq2Bo7RwNEMKorWDN2ZPUhjqZDUaeoiDjdx6+dYPX8/q5dOEOoNShOxBRgT0DqNoyo/ZKauKE3KwBURY1XOMoaisBlYSd6GVhX4RjLcUFw637CyWJM8+CLtvqTse7o9jSmgKAK2DEgNSE1Vx9Rda5kW/o0n+hwNWgfcOFJtevwYXJOs73OQYequg8PjMCXYlkDY5LknZTKyjU5RjSTDQcpIIVs7qV6fqb2PY+HIDbSmryLqnQTRyhnJ2cMkRpqmyWN9VlPIxDn03uVRN+9FCYiQyMi+jrhas3Rugw/8zYf52Af+icHyBkEo+nv28JJfeRUveeUrmJmd/a4Or6lrvvDpz/D2t7yFe+/4GrKpMFpy9bVX8Owrx0LqAAAgAElEQVSf/GEWrtxN1B4fPaCRUW8DFCE6UHl3JxO5Sck0sUSZSTwh7fVcSETupPqJGTFXOaoy51QTUEhEozhx9yk+9oFPcuLBR3A+8oAPl4H/78HH97zo/fFbfzf+z//xN5x56CFE7fLuTlAAOwrBtQd7HNtn6Ygh2lXI4MhLoCxdSpR7k3M+G9cAZCtscpsVCULhTMFIdHj4nOM7JzZYHAeqCA0RJyRYza6DB3jhz72EF91yC7v37ku7PhKY4F3DmUdO8v73vIf3vee9LJ8/h/AeFZPvQKkke+csT79phu+/uWB+1zpWDlDZnSXIBDYkKk5gXAmWtgRyxxVc9bTnUJseo7ohhoroG2KA0dAlWshwjHcNdVPhXI2MDd1SoUJDtXoOsbWMrNbpyhqrAi5INkOH3v7rWXjczXgzw7gZEeMYET2uaYhR4oTCxVRghGgyBURilaW0PWKwxGCIXkIQhNjQBMc4BGqf3resx8jxJvXWJdaWTnLp/ANUm5do60C7TEFESgeszaoLo5LhqIpYK4gSVKGyoN5jjE7KDJWyI0pboNC4KjIeSBbPO9ZXAmurDVInO/upWShbnrItsKWg1dPYlsH7bKMlk2lpXVc0VYOMCUl248h46BltBdxQEF3ESIVWBikk3geapkZoUIVAGEXMEZQT3qJzKTTde4kIPXR7D/NX3UBn3xMRxW6U6BNjkcYaERL9J1y2A/t/AkAwGd1TGl1q9xwieGjAjyXL5zd5///4IB99/0cZLG8QUdjZWW659dXc8kuvZMfOndsFzxNxdc1XPv8Ffvf1v8P93/gGsqnQWnDl1Ud4zvOfxYFjexDWE5XHR5GdtVWK/8xotlCJthRFdh0SAZlBEfISI4Edadcp4qM5jrDtgxg9SidxnXCKUCkevvc0H/vQpzjx0Gnur/33tOh9z3d6P/ujzz2+ceEiym0nKdDVkiOzBU+5ssN1RyW7e0M6sqJUAm0TTyFksqsxEqtNtpoPlKVFKlKmhEjUhlapaFnoFo7pnmduVtMtLb4WDOvI9sHiA1srq9xx++3cd++9zO/ZxfyuXSidaDRSKaamZ7n+hhu46pprOHPhAucvXcT7nNURIlvDwInTYxaXI91en07fYE2SRomYIwuNIKCovGIrtJg7+mT6C1fhRZFpE4lSUo89TR0ZVw2jUU01HDPeGuC2BshhTb2yyurJ+zGDi7TCFl1d0zY5CJqIE46hr5ie20WrNQ9iiii6BFkQhMFLjdAWYQqkKlGmwBRtlG6j7RTKzCDEFII+kjZCF0RliMKAsgm4IK+ipEHZDkXZoyxb1LXDuUT01YCSoDOaKqTIJgE5AyOkPWsIMWtySWTuCeE1JrOAGBOVpbDp1zWuIhsbgWpoCI1Koz8pTyQKg1IGhM+dCQgRiDHtAfWkACuJzyL9mBUzUSROoM5hTUan8HSlI9pmdxgVsSaZCdgy0uoqWi2BsIbW3D52H7kR3T5KEDvwQkFsENHlUVU++ha4PMpO/k3IKGuSfkysMQgSnGFtacQH/+4jfOhvP8RgeZ2IppjdwYtf9Qpe/su/zI75ncmFhvQzrOqar3zxi7z1t3+H+77xdYSrUVpw5MqDPPt5z2Lf4Z1EVROVI4iQ0djEnxQikfCjSBGlMTZpUsERQk2IVUb7HSFUaUeZnyGN7iF1dsKDcJcnMjFhC3hi8ExNTzGzY5alpRVu+dVff2wDGe983euPpzAfiRaK+Y7kyVd0eeq1HQ7sHDJTDmjLiJY6WTSZSFAp/9OWKo8JEmMSsTVzJDCFRAiHLQVlW9JpW4wUFNpj9Yh+X7Bn9xStQrM19AydyOHWEZqG0w8/zJdvuw0fPEeOHaUs2zlPQqCLgoPHjvLUm25CasNDDz/EeDRKvLQoqBo4e7Hh4YfHSNFj5+5ZyjbI6JOZaQi4ENisLY3azaErn45u7yIEg8QmQmvtGQ4axqOawcaAwdomo7VN3NqAsD5gdGmF4fnTzKgRM2WgX3raObEscnmfFYSnnJ2m1Zsjqj5BtBGqJGpNlCoVK1miZRujLUpotGqhdRtBiRRtYrTplBa5GgmFEgojVSL5CkWQhiANUhdY28XaLtHDaFgTXEQLMEohtco85bywR6SsVpJaIsncoGmym8jE3QWSKalzlKWl3bPYQuGdZ7jZsLURqStF49O4LATJGzGErHsO+Sac7A3TrjFB6QK8JFQCX5HMCiSQ9b+JYxlRFnRBju2MKexJR3SRSNO2FYnWszHa4tLiEC3m6JW7kWhQFSnoPNFvLhPEL+/ytm3+xWTEFlnX6pN8zms2Vkd8+H3/yAfe+wE2Lq0BEjszw4tf9Upe+Wu3smNuDi0nHR40zvGN22/nD//zG7j3K3cgXY00giNXHeS5P/XDHL56AV1GUD5b5SdbsdSR+lSsJvxNESAmilF6XU3+nMzk+UR6nEgmY/TpZ5+ljyn/I615lFKTgQ0hI8ZqpudmmJuf46Yf+onHdtH70ze88bgQEltKrthX8K+/v8/Tn1Sws79CW41oGUWrU2A6Aq8qgnT0p1vYrki/KJlRTgnIQBQO046YNnSnLb2+JYgKU3raPU1RapQImGLMzI6Go1f2mdtZMBwH1keJb6WIqBDYWlnh9i9+kYcefICFfQvM7phDGZPOKimZnt3BDU97KkeOHeORU6dZXVklOp+s7wOsbQYefHjA0mJkenoHnZ6mUBEbBY2PbDpo7TjM7gNPQdqdRGkAnfhs3jEeVYw2RwxW1hktrVItrTC8eJGNs+fxm0vsnhLMT2taqsYojzZpn+Qz7SIg8NERlaG/6zDR9AmqICpyBKXAqBZaZjcaqVHSpM+pLlKUKGnTKkEnHtu2kSdZ4YACkYpikkpJpC6xRY9WawqJpq4Sidm7OvO70lI7ipgQZqXTzZYVJz67wHh32Qcw5uVqSkdL45LNXZaUkbIokbLDYBAZj/MSPTRoCZLc6hGygWlKHNt+OHBjyeayx41TaI5W5HGbTGUSSJNADV2I7AUYs/Y08dWUkqlICs/60hJnTp5nZmoPrXI25WxM1ApMpGzfXey2s2IntBSZCoUMAhk0W2sVH/zbf+R9f/V+1i+uIqKkmN3BC3/h5bzy1/5d6vBywQtA09R886t38Ae/c5y7vnQ7wtVgBIeuPsTzXvhsDl+9AKZBWI8wCqUNSmq0lknnrHPWhpbbZhAy238hUmi71WDUAKMcStl0gMm0wpA6yT3V5HmUxmiTn18lQEQmbTkySUlnd+7gqsc/87Fd9P7rW950fG7e8syn7+AnnjvDddcOmetv0CtcsmXvarypwTqiiigrafUtRTvltRalpNU1SJvcevuzllZP0umZtNdpq+R4bAKy5RClo+wrOn3J1HSgP1ez57Dk2NVztFoFa5s1o1FC1jQQm4aHHrifr3zlq0ghWNi/j3anncYPIbBFwRVXJqPSIOCRR04xGg7zOAtVDafPjTlxckQQbfozXcoWVMGz0Sh27L+S2V3HiPRBlLhYM6oGrK2tsL68yubiCoOLy4zOX2Lj9Fn8+jplrJjrw+y0olU4WsXkpkxKBCUEEY2TCm8KvOnRmdmPbu0kKAsqInVIJqSiREmNMQVKaqS0KFUABoFFKPOoZDWVSM/K5Is7hdAIIXOYTSI7K22Q2iJ1i6Lo0yp7NC6wNRhSV0kjOgkqTnyvxOvymYsYfRp3J138JOCcSdbsJO5RpUhLW0aEcthS0+vPYOwMzpfUDlzWqgZHKtAApIjPpkp2UzSGjeXAxkoCObRWRNy29ExIkEZsp6RpI5N3oQAm5Nzsc6e8oJXW9mxsrHDu/EU8lv6OnUiVujyyP+FE8yylTNbxkSSZA1BZUx5Aes1wvebDf/9PvO+v/p7Vc8tpupma5qde+nP84q23Mr9nD1qq7YJXNzV33vE13nb8jXzrC19ENBVCw5FrjvD8n34Oj3vCYUw7odNCS7QtsLZIPoU6xXwqY1JYuNVIkwufTUVMK01hDFZtoTmDlkNs0ceYHsJolFUomwAkbTTaWIxOIeTGmPR9jEZamxgFWmKsxhaWo1fc/Nguep/9uz84/oKf2stzntPm2KENpnsjyran1dfYviS0arARaVQ6YW0EE4naEVWDbQmKLhRdQdkXFL2IKgIxn2Co5HhiOxLdEaheoJj19Oc13R2a1nSgPwPzeyNHr+qza1ePzdWKzY0soicpAZaXlrjjjq9w8sQJ9uzdw86dO7elWkJJZufnedKNN7D/8CHOXjjP4uLi5a7Pw8qa56GTQ5ZXoDPVR3c0jW0xs/9aOjuOEkUbH2BcjxlsrXPx3FnWLyyycfYiaw+forp4CTOq6JvAdCcwM63odSNSOdo9g7YCIRONAgRRaLwIeBEJqk2rt4+iu5cmWgIkPh0FShaJziFMpnUYhDCkUJxJhzQZMWVSfWRliBAqC/ZFDuFO/LwgIEoBUiN1iSk7lO0uQmiqusbVdSpckwBtkfZYgZjiIRE5mzVuW/cLoYguycoECmNSx+B8Q1kqOl0JckztR5TtNrv3HGVm5jCBDlsjn8mzMTkce/AOXKOptyz1pmXlYk0zzn7GeVEZw4QoTB7NkiJEy6RRnoisQkxZICmkSaJIUjakZ3V1wKgSLOw7ijKdHNOYfp6KlGQXZbLrEiEHEcm0VxRRohrNaK3hH//nJ/nr//63LJ1dhCAop2d5wctu4VW/8evsWVjAPKrgNU3Nd77xDd52/A186/O3IaoKNBy+5gj/9qUv4LonX4nNBc+WFmOL/NCYQqcUNVtgrE16aGvQVqNLgy0s1pYZ6Kox4hw0D6PFFmWrhy13oIsetlVgCosu0nMVtpVDioqUx2tV/l6pINrCUhQWbSQHD37/Y7vomZN/efyJ1zhmuqtYPUDqMaoVCIVElBHViqhSocu0OylaCtsW2K5EtSOqFSi6AtsF243YDlB6TBfaXY1tQdEB043YPpTTitaMQHUEsivQnQahGmR0WGrme5qFmT5xABtrDZWbjCKRajzm/vvu4+tfvYPCWhb2LtBqt5lolopWmysedzVPfuqNjJuKU2fOpF1flmKNx3Dm7JhHHqmAPtN79rP/cU9A93dRe0UTxqytLrK2vMxwZZXh4jLLD53ELS7RA3om0O9CpytotQJlGWm1ZXYsIYnxZUQaTRAOF5o0KoqS9swR7MwhKtEC0o1LTJ2dFBbQqMnfoyRB45PwILKnW1J9MNnDCZUVHHk8k5f/nYxBZXKwNhZTdCnKLkoZtrYGjMajTBTPaGVegGd+BROrK7IOVWYOWIwgpQEyBcU7hPTYQtAqFVp5XDPAB0+7M8fU9BG6/YOYcpYQNXWtqKtIjAYhOwy2NOdOjNhaTXGKSid7JLhssCoArVUOEpfILMUTQuWCnVHYbN5auRQLoNM2j+GgotvbSX96PoNIeVEoJCFqiMkIISII+Bxar7ChZLhc8bEPfIL3vPuvWT6zCEFi+9P8+M+8iF/+jV9n98LCdocXyQXvm9/k7W98M3d8+l8Q1RhhBIevPcbPvfLFPOmmx9PqSlSRuI3GFCk1zRqMVWhrMEUqgnZS9IxNxcmqVAhVgVUj/PgBBmvfpF+so8SQrcEGZbtL2Z5DFz10YdDGYEyZil4upNpatNUok/iq1lisTsXVWM3ePTc+toveqQ++/rh1DuUhUBN12jWl0AiFsBFVRGxbUnYUrZ6m6EmKrsT2FK1pQ9kzFG1J0VWUfU0xrSi7irInafUErb7E9sD0BKobUS2ZkslkNoisI9W6oF6yjC40NKubdE2gqw31CAZ1xOebT4TAyqVLfPVLX+b0I4+w78B+dszvTMtfslHp/DxPfupT2XfwAI+cPc2lxUsQUyhRCLC86nj4xIBqNM3hK57A7M55Ygxsrl1gc3WRpVNnOXfPSZYfOE2zuExfBgocva6gO6UoOxFbBtptRasl0TYiTcDYZF+PUcmQNQRcUAQ9zdTCddDdRxRt9MS5NGqIJnPecpeXPUiEmnRz4nJhywFDMYLI73diqSGSyBWkyvutJPyXWiOVRSiLNi3a3Wls2cY7T9M0NON6O7VtEtyNiJedjbMKS5Ks4dMeLjdSMo1ERaEwKmAMlKXGSGhGmzSuRuserdYelJ5Ht+ax7V1E0aYJBYgutW9xcXHEYNMjAsnmivAom/k8vouJFf3EYDU5pvgm6UyFlMnoVqRAdSlS/I2WktHIcfHSGvN7DqGLLi6mLpG8wRMT6ZsELyISjW4K6g3Pxz/4z7z3z/6ai6cuIIJEd3r82Iv/La9+7WtYOHDguwuea7jrzjv5oze9mTs+9RlENSYqOPy4I/zsK36GG27+Poo2SBsSNclYlDIYrdBGoQqDtgatVXqYVLS0NtiiQCuDVSUtXSPdCdaXb6eQZ+m3a7SqqetN6mbMzPROynIOpbupyytaaG3R1iSDWS2z0Wz6XkalaFNlDMZa5ue+77Fd9O76i/98fHOpZjwKxNAmxhIhwRQRZUuccCgdkMohlEfZgLCBaAJojy4UupSoMqJKkC2B0AF0jSoCqvBI6xHagQnJ0FNJtCwRwVNvaoZLPTYuWJZOVCydGNFsNrRsxd69iiOHptEo1jcdjYvImI1Kh2MevPte7vjqV+l0O+w/eBBbFNtdX9npcNW11/Dkp97IxnDAI6dOUVdVtokKjMee04+scde3TlPYDnt372JzdRG3tcmDX7+TU9+4m7C0zrSQtBUY4ynaHt3y6FakbAvKlsQUHm0C2gZEIfBG0BiBQ1HXksprYnuO/t4rkcVORCi2kUERUrKXlCZ79U26l4QeSjGhXk/U9/kjv8dJMLjUOnd5SY0wIWILkUABKW2K2tQGbUta7R6tzhQxSAZbQ9y4hu00r3QLC5G6Jy00uLiNuopt1Dct0qXKtJk0HCNVSNQSCaIe0lRDGi9QRZtoWgTVJehphJ4miC5CdLDFFF6WjMcVrq6JXuBcBk2yfjiRhBOVqHEhuU27kPiLMfHSMixLchaG1CMqYtQsrY6xZZ/ZuXlE3u2J6BOjLa8DYk5aU87gt+ATH/pn/vxP/5ILJ88igkK22zz3RT/Nrb/1mywcOLA90kag8Y6H7r+ft73hTdz+yU9BNQINhx53mJe9+haedvOTKLsir3zIBUjnMCeJLQuk1hhj0WpS+EyKG8ift8pQ6Ao/vo/l819E+dP0ygqrHUp6lAj4ZkTdjOn2Z2i15jC6hzY2Axm5uKkEmmhtUEZjrUFJgTIaYwyz049/bBe90x9743HnYXPVsb7a4GpBsr9V+GjzvqhBq+TKK2TiHkUVQUY8Pk1iGoSKBDw+NimfVSVullLJf02qREvxTUEzsFSrgc0LnqWTcOHBIYOV5OvW7kSmegU7ZjU75hsOHimZnemwNYgMBsklRCGRPrJ84SKf/+xnWbx0kf0HDzA9M4NU6WJUSrJzfhc33fwMdu7ezcOPPMLaWrKnlzHinePSxSXu/MbdXDh3ASstGxdWOPHtu/Grq/REoGsiRY6cbHUEnb6i3VO0OgpjPa02aZFfCkIhqEQApQlA1UQqL6mVoj0zjy2mUreS4rcSkhqzBbrIKGTu7pITTNKcep/7EilTUYoi5+2GrBnlss42j7Yxj4hKpXFZiOSSHGVEGYMqSspuH20KxsMxrq7ABaIPuasSqCgTGddlS3kREUpgdL7ZRUQJclZE3HY1VjqictiQaIZsbF5kdbiGshpbljgs6D5R9ZGqi9FtWp0+uigYVTXDzTH1KHV7qePJBPUcsZks5iXeTxym0w4yeCAmykfKBiZFlAZwlWNxZYvpnbspu30mcZpM3HRE8vxT0dIMHJ/6yKd59zv/jAsnzqbdXrvLs1/4b7j1t/4j+w8eRGficCDRUh66715+//gb+NInPkEYD8HAwasP8fJffRlP/1c3UnYE0jSZTT6JDUivcYKsam2zT6POf6Y9rVIKrSJGbVJtfZvzp/4Z6R6m36ooDai840z0nobxeJWtwSrdfp+yPYOQRQbAFEbrhBJrvV0AJ2dqCrRSzExd+9gueuc/97vHRZkyRYWPbK03rK9FBlWP2s/jKQm+QueLJPq81GbC/QGjVSKWxiQ8D8GRwiPTsjj6lI/gG8Ngq2BzY461i9NcesRx8aEBw0WHldDuKYopkdDfvsQWjqJsmJptOHC0ZP+haZyHlQ1H1SR0VgF+NOaeb3+Hr99xB91el/3791MUZnskbLXaXPOEJ/DEJz+ZzdGQU6dP01QVkhRCPdqqeOi+M3z7zgc5f/IcxgX6hUg7vJam043YtsO2BGXb0OoITNFQlIGiAF0ovAWnU8cg0eAljYMmQLQF3Z27kcriXCaGxpSiNUER02GRRzUxeaTuTuab5NEBMElTC5fj/0K6cGUiIBPJ4eUp4zaVjdwBaZEs8K3FtjqUrS5NE2iqGj9uUg6HC8ggk9g/pOedhKJLmRBdmXMxvPOEkJyMhVSJ06sCyMQjkzpA2GI0XEUbSdkyuaMtEbqL0BpdFpStKWzRIyqVrJnqnGciEprsvc85IuTMDYF3ST8dXCTWApzKhdEn6yudxmKJYLC2zlbtmd5zFGlaRCQeiYsCokIFhRs0fPojn+bP3vnfOZ8Lnmx1+MGfeB6//tv/iQNHDmOkvlzwvOPB++7lbW98M5/76D8SqyHCCA5cdYiX/uotfP8PPZWyJ1E28/DQSJFiQ7VOB5lUZrvzSqsOkeMEkoWXVA2KZXx1N4tnP42JJ5ju1pQ2xQKk9QYoHdJEUnicX6eqN+j2eyjdQuoCrXQ60HJgs1R6O7FP5EQ/rSX97jWP7aJ38bY3H1dlMtw0RmGUwQXNmfOBu+5puHAecBIVNcGnzAMRUuaD0mw7WkiVMgzSCJRyUUXQxEbja8Vg07C23OL0yYJvfhXuvXPIeH1Ezyj6bUW7D8VMpOhETCmwXYHpOmzHY4qALWtm5wMHj3WZmipZ33BsDTw+JLNM4QKXzp3jy1+8jbW1VXbv3ZONSpM/k9Ka3Qv7uOFpN9Gbmeb0mdNsrG8QvUdGCC6wuVGztjICDzumesxOQacdKK2gLANFO9DqknSgZaTdSYFB0Up8LnhaKISX+EakpDSvKcoZZuYOgJnChyJnSsB2KE5qo9L7iFzm3MWJnhkmc5gUGkh7u0QsnRQ7tsdbYjqYti9yknlA8A0ZpdgORZLaULQ7lO0pQtQMt0bUWw2xTkBCep0KJXNkoJgUnxTiE1wkhETolULhsyGoUIlLZ3TEKigMGFFRD5eoqy209Fib95hSIq1EWoPSLdqdWZRtUdUN1XAELqbDM6S9Y0pQS6h89KnDEyFzc31ImtwwyRVO+cZpXJesbw2x7Rk60/vwwuKFRgqDiZow9Hzu41/gL//0/+TMA6eQeYf3rJ98Pq95/es4dOzYdxe84Hj4gQd45+/9Pp/98Ee3R9oDVx3kZb/yEn7w2U/HdhRCh8z8Sci7xKRdoJ6g7ol+JDIwpVSOFBWgZINSl6iGdzFcu522OUu/M8Yqtz11+ZiUFZM4AFtEul2JNg1VvYkpCmzZRogWQhiEUmn9kfmNcvuROJvd9lWP7aK3dOcbjgs7QcEarAFrI61ugdIt1ha3qLfGjDcjobb42hJDxDmHIBe8SYgKId1wXlIPYbwRGW9ZNhZbLJ9XnD3RcP5UxeZKYNdsyd6dgal2RdHyUDbojAqbjqfsSUxXIdsRrEOamBK9WmMWFjSHD80SgmRltWbcpI5NRagGA7595zf59re+RX9qioV9+zC22O6c2r0e1z3peq67/nrWtjY5feYMrk4mAIRA1QTWNx2jzUC322Z6h6JsNxRW0u0Y2t2AKR1FKbGlJuqIUwG0QMlU8IKDugkMRpGxK2lPH6Qze4QxXcaNonES50Pq2pRMo2lGmJEKL9KGjIyoxuzoq2NER4UWFqmSB2IMAUkqBmLSEeZw7mQ9TzqI0u2RyLb580KA0gk0Ubak3ZnB2BbVuGa0NaIaNvg64P1EupXzGEJylQGJayIEiURtE5qbJts1SVBaYq2g0IJCgZEe4TYZbi0TvKMsOomcLW2iiihQ1lIUPYpWn4jCjRtCnTs6f3nXNxlzRZSILK6XMnEJg0sdoZYaKcEIhRYlm1uetTXYs+dqjO0ThaaIhjhs+NzHv8if/8lfceq+k0m8X7b5gR9/Lrf+1n/k6JVXpkMkXeW44Dn18Ane8dbf41P/8CHCcAAKDlxxkJf/yi380HOfQdFRROkSGC81MXsECgEomQqg1JmHmdH6bVWISPtFMcZVD3Lp3OdR4QRtu0lpQnYdF5lfOOEvCrRRCQVWEagZVwMGg4pudzdKzRNpEyWZ/JhoUUJMip8BYei2jj62i97yPW86roxGK5H0jTagpaewjnYrsme+pG1rRhsVqxdrlhYrXJ1S6oNXiGDTbiUk++96pKi3DMM1weaq5NJpycn7h6yer4mVY8e04sBCm+neiH57gLUNthspugrbUZiORHUithtQJQQt8BKCTlImiacwDfPzgmNXzNDrlKys1GwOPSFKVBTQOC6cOcOXb7uNlbUVDhw8RH9qOhlqAkob9u7fzw03PY3e1BSPnD7Nxvp6Jq2mTmV92HBpscYHw9Rsm96UpCyTgL+0AWtDIsZqjVCTiy/5vLkAVRXZ3BIMmw5zB67DF3NsNFB5j9Qmo6wCLZP5gcwXcB1ccqaPyaE6Ro/CYZst9HgVUw8xMeVpkGzeMro7GYcVEzNOpdSjPPLSR9oTpV2RlimgJyIQyqB1QavVodXt0HjHeDSiqRyuClSVZxLoFIPAe4F30NQZVPCR0AS8SzpoQeoS042dMoa1BBEcVoGiYbC5wnC0RavVotBthCiSXbn+X+S9+bNu6VXf93mmPbzDOeeO3ff2vT2ouzXPatRqDUAMAgvMoCpDwCmMDalQhkolBFmxA0ItjImNI5ySDMJmFEZFwERAOUCAEDFIQqi7JVpqSS311Z2nM5/zjnvvZ8oPaycDnswAACAASURBVL+nW/kDSFXnVN2urlP3nPtOe+31rPX9fr4KrMGVIwbVBrao8T7SLhp8E/piR88CzP1IQAqHKwtylrCi2FNXUo9Tt8agkmH31i4HW1NOHbuT2g0Iy8hf/dnj/ML7PsiVpy9CUFBWvOlt38j/8GM/ygtf8lKctUcIeR8DVy9d4t/+5E/xpx/+XeJiTtaJM/ed5fv+yffyd972ZtzIgIm97Mj0BQVWNk1Q8lyN7R0gclPRui9AOEyOGDaZ7j7J9q3HsRxgjdCNJKtldRJQfYRnP/dVlpQSbedZzFumhw2j0R3U9R2g1kGXos7oZ7e6L7orGMewuu/5XfSmX/qpR43ROGvEImRBF5KToFXA6EDhPIU1GAqaZWRvt2Mx0STv8I0jRUfoEt1CszysOdw07N3O3LoS2breoIJiPChZX1OMRp7CBcqyox6Inq+oNeXQUQw0tpTjo6nkcSgNMWqy0mSVUEY6G6MCg0HL+bsrzp9fpw2wt9/Recl20SmznE757JNP8rmnnuLYsWPccdcZXFECUiDG43Ve8ZrX8NJXvIKdvT1u3rqFD724OMOyVdzaDuzvJUZrA07dWVEPA1Ul4TjKJpSDpB0o4fOFJL7V5TRzuDBU63ezdvp+mmzxfZ6ByQqbNaUtKK0VT6zOvasioZLQbnRomexeY7l9kenVz7LzpU+yf/lJNq9+nvlsh8GowpZlj1juNX05HRUAEDN9Eo0Lxlh0f+RdbTbphchyfARlDLooqMdjTFnStC1d05G9dJ4pR4JPxNCDYUMftpP6YCMUFgUrh4QypKzldc2QYkAhtkWrEmF5yGy6j9KGQTVCq1o6IqtIRveuknWcG4LW+KbDLzsJGc+ywMlZLnzJHhGsXEjpCKXuY8SHgMqJgozxgc2rN7n85atMpks+9+SX+dUP/CcuP31J5Dn1gDd841v5p+/+cR586UsorDva0oYUuX7lCj/70/8Lf/zbHyYuZmSdOXPfWb7/B/8xX//NX0ux4YhOUFAaOUaK0b8PhO9vUquv5wJMxdcp9J5Cb7M4eIrrlz6G6rYYVqlPocu9n3lV8PumrUf5i1g7Q9RED+0isJx3jEYblNUaKddkrDSWWRZQR5+VlBjVL3h+F72Dz/3koymmfjSexNSvQTmFK42go1QnCVjaYnVB5Wq6JrOzuWR3u2O6n2kXMN1T3LzkuXmlYW+zQ0fH2nDIsLIMRpl61GFKTzmAaqRRZUY5hS6VYN0NZBV7b2HuiR194HUvyEWBtglFxOTEwHnOnMo88OAGg7pgd6dhuhSRl8qJ3HluXL7Cx/7iL5hMDnngwfsZjdeOAo6dKzh3zz284S1vZrS+zjMXLjCfz4+sV94rtg/g6jVPwnHy3AajDYWyHcbJ41Y2gylI0Qim22emc83Cjzh57sVQrbHoOhkJpCzK/6gxOuNMxmlFYTWFTpQ5kRdz4uEem1/8DI/90e9x88mPM7/6OZabFwiHl5hufomdWxdo2wnHjx/DupKchZGj+g5S3s++o+iRQsY40fipZ5OzUkacGUFEjJlMUqCdox6NqYZjcaosWkLrIWRCzASf+vBx+gtPNsfOaKFL9/M3lSF0oS9Gut+6RozJEkpEhNAw3d8m5khdDynsCChQVoofxmLcgLpaoxoOJHe2DUS/cmHQY+azhIP3YUOkVeIahGUCn1AhYjE4M2R7c8JffOSz/OHvPsGNy/vkaNBVycNv/Xre+Z5388KXvfSo4IF4gW9du877/vVP84e/+Z+I8xlZJU7ffYb/+oe+n7e9/RsojxUkJ1s2g0Fn6Q9zFgLNyhu7KnTPhR4YbSBZLODYo5s8zdULf05YXOb4BtQ1OKt6C57MclUvL1IyFu67+izzQQxWW5y2zCcLmsazNj6GK9eAAZLx0s+BV/BUYFQ9z4ve3pP/8tHo++xOMmgl5nanwYLWllUcYY6qF9UmnFMUhSF0iYPdzNbNyM6tyHKiKG1mfVxQlwajE0UBrorYKmIrTTEo0FXPR7M9a63PFCVLylNMma6fe1lbyN2MKHo2bcTrqiM2ZoyKlIOGO85X3Hn2OLNZ5nAqXYjcFBPLyYTPfPrTfPELT3Py1EnuPHOm35YBRjNaH/Pqh17Li17yEja3ttjcvE0MEhSTEsyW8OVrC25tRerxOmsnSmwVyYS+QAcBZLaa+RQmM40bn2Nw4hxNyHQ+CGI+ZFIWeUFhIi4uiPvb3L50mSuf/wKXP/MZnvroX3LxsY9z/aknYPcWYz9ngMeoQEwt64NArecsD2+zmOwyGFQ46zC6QKmeDNovHZ6V+D3nIlsdZ7TBaoPJK5pIgBVuCEja4KoB9WgDrSuWS0+78OKj7cdqz8olevN65ginDtI9qJzFGiYsl37umLBGSClaZQyBdrbLfD6jqsZ94StJKLIFXRpcWcusbzAkGo1vO0KX+uP2KoqgVzZm3bPlggSPt5rcZGKnaLylCQW3Ng2PP7bP9q2WGA2pVLzma97Ej/z4u3jJK19JYe1RwQspcvPGDX7237yX//yh3yDNZ6ASp+++k+/9gX/E277971KsGcKK5pJApb64aTmGK0SKtCp66jntniyJHPJx3Ed117j+zMeYbH+RupgzGAaM7frITnmucmM/kljTG6T71x36TQ8qKbplYj5pAMv6iVMkUwI1qJUDJh99Pobl8/x4u/U3/+LRGPuAEQXK6J7UCm0XCTFDdjQNsk4nk3NLpkXrRFlYCqvQuqJ0BevjxHCgsCagdIOxEVcqTKGJKuPqkqgiiSj0h152EVOvW1OarCDoKFaq/mKxRrZaKcV+tmIIWbOMiVnIHLSR1mlO3HcHr3rDQ9xx5gFubR0ymUzFZ0om+8DVixf52F/+JW3bcvfddzNcGx/NxowruPcF9/Hwm95IURVcvHSRxWKGQn6+85mbtzsuXGoIacjxkxsUVSanjtgE0jziF4nlsqBLx9i4436iHTBvOtq2E+R7l4i+pUgdZj5l+/NP86WPfYIrj32aK5/9Altf/jLdrVuE/W1sO6POgu3He8iRorboUjMsodJL/GKb/a2rTA52KJ2jrkc9+NL20gSRFX3FdlepPlgn9WJkeT9SD6ckSSDS6lisraOoRrhyhPeZ+eGM2ElWibDnpLOSQqik2woJrSSESELBk3zGep2hCK5zL9uIGJspi0xsJ0z2d3GupirHKCqULsTBU2SwWrzE1Tq6qOQo28pIgj4NLMcsOkaTSFmROktsswRz55KZd3zpcuLP/3zC5o2AT4ZUKB76L76ad777Pbz8NQ8J1ACZmMUUuX3zFj/3Xil4YTohq8Txsyf5hz/wPXzz3/8mqvWC6DqSiqvSDoh2lX7u+P/m9sn7QS8yl7GDUQt03GTzyhNc+vwn0PGA0ShR1QLClS0/QO63/T1tOq+OzLkXXivSChQRNV2jmB207OwcMFjbYLh2HGVqJP0m9zZE4P8PRe/WZ9/z6OqYk4EQY5/dEGnaJNSNrDDGYawiqyBgR6f7uU3C6MxoaBjWGa09WkWcy1QDLfO5UuEGBlMako5EMj4EUhY0dkzxiPOvtCLbLLYcbfo5Q+zFuBr6LIQ2ZqZtyaFfoytOMLrzhZy8+9Wce8nD3POKl/Gy172C8/few+7hPru7B4TezaFSZn54yKcee4wLzzzDHXfcwek7TmNdb/LXhvVjx3jo4dfzwIteyO2tLbY2N0kraUvKHB5GnrkwZ3sLRsNjDIcFOkWyVyyXmrm3FBvnMeNzzBpomo7loqOZe8JiiWta4tY2l/7qk+x9/gLNzV3S4RLXBdZMpgyBKkZKBc4WaFOQEzijGdYOhUcVkLTHqhYVDlnOtjncv82gLqnqEeiCtBJCE3rGWv8aH3V+sl3NWkTlmUjISTIWYiSHjuAbQgzElLFFwbAeEbvAcrYgdxEd5ZiaszgkiKCS5HM4LeHRRqkjEGmIAjbQ2vSFIGE0FFbjbMI6j8pLltNdQDMYrImjRBeCSzcerKIoB5TVGkU1BGMJIaGVw9gSyEQdiBlUtDJKU4loCpo85pnLhj/5yJxr1zxtssTS8LI3vYF3vvsneMXrvgpTCI8YpMu/fesWv/j+9/PhX/01uoND0IkTd53ie37gv+Jbv+tbqNZrspMbhs4CclAAWth2guaSYqefMz9TSh3JRZTSaCKOGYvtZ/jSp/6M5d51clxSDaAaKUyhZP9AH5HZy3aCp4/E7N9bJZh7EbdrYtBMDwLTvY523tEsAidOnaOsj5FVQVZaMjiUIPpHz/dFxtZnf+LRkAEtgTjCJ5O7ROiHrolEVh50JGnJji1LJz3YqovQSe7YJlOUirJWFCNLKgLZRXSRSMaL/KJn763YYNoqAVsaQ5LTq3Rm/cAoovBB0baKpskcNplZHqLW7ubU3a/n7P1fzdkXvJ7RsfsohudoGWCs4uxdJ7j/wXM07YKb17dpfS+gzpnYBS59+SKPP/Y4wXvOnTvHaDyS56sUtii57/77ec3rX082mqvXrjGfz1D9DKxrM9dvdFy51BLbARsbaygLM59p7Trl8btYJsNsnphPOtppS5guCPsH+NvbTL54iXRrFzfrcG1G+YQxShZGciLCKUeIiraTQXxVWsmgqDOuzuhKoUuLNYnaNhAnTA62Uc5SDEZoV5HQhBSQOY9GyysrMpYeICyW3kQiEaJAEkKI+GZJO58ROgnMzsnjjGI8HOKMJUdPaTJrQ0fhxDqWkyDfVVLoJAsNH/q4xCS6OqWMCLHpeYA5YzEYZVEktA1o09Eu9miXU6pqQFkO0UqO78opskkoayiqMVW1jqsH6LLEDmrM0JKcQSuH1QqtAlhLl9e4cLnkI3824+rVjiZafGV4xZsf4Z/++Ht41Ve9HlvY/qYgQUWbN27wC+97H//7r3ywL3iRE2dP8V3f951823f9Par1mmgCqIRbdV49i09r1TtwpOA9N3QohPCcY64WKGxuUYtNrn7uYxzcvkBhO5TpMKWlrB1lKQqGFCIpiFogRY1v81GWcc65p1yv7HuG5aJgNkmELlOYitAmjKnYOHEWZUdk7Y7MBSllxs/3RcbuZ97z6Cp0mixhIynLVCep3B95ZamQdEIVYjJfGdPF+2l6MapsYm0JtlKYWqFrTS4yWfcUV5vBJCl6RjDmxvTDXi13Mafl+1llApplp2m7ktlSsaTGHjvL2Zc9wpmXPsLanS+kGJ0l62Og10jR4nNksdhne+sSJreMCsvB9ha+6Wi82NjkKJY52N3jU48/xjNf/CLn7jrLqVOnMM6SEd3diZOneN3Dr+f+F7+IazdvsLW5RYoBg4KomO3D5S83bN2OFMM17PqAtTN3YurjzOaR2bTlcL+lmUxoN7dJW3uk7QPMwYyyixifySGT+iNfDgmjNM46SEqCrIMkjlW1oh5AOcjoIlHUDu0kW9aYQKE6TJxzuLeJ7zzrx+5AuxE+aQE2aHp5hGVFJtG9Mh/ZGxNTJMZI8C1tM6drF2Qf8E1LaJeQPFVhcDqR/YxBDRtrBcPaoBW0TcT71BOYewR9eK6LQrqSFEX8nHMPA+2pvkZ2vyjlMbZjudxjPt3DFRVVOcboEqUkIF45yfawRU1Vj6F0+KLGjNcZnzzLcO0YupDFTEgjvnzF8ScfOeTKlZYuOUKpedkbX8873v0or3v4DRSFRqkV4UWxv73Dr/zc+/ntX/0gy705SmWOnTnOd33fd/D27/5mhieGeB1gRddJmYwha/scevGzBe+5mbrW2n7K11Ork8f6XW5+6WNc+txHSe0hXbfAp4CxitGooCgT1sSjmSXZkILIo1KkJ1KLRXAlbk7BsZxp9nc7FrMoaK+UmB3uMxiOGK+fJquSjDkKMR8/3zu9g8/95KMrBcMqk3ZFJe9HNCIvcFrQNk4IELlnmGXdz81leYgq8hFzz5QKV8vSwTjJRV11ekZDYZRAW7WIK41ROKMojEYZxzIa5q1m0ZYENWbj7IPc9eLXcPqBV1KfuptcruOzI2ZDUopIpvUds/khu1u3WOzvcfviBS4/9RS6mXF6o2ZgNV3Tdx990xnajksXLvDJT3yClBP33HsPw8HwaMNWVjUveOABHnr4YZKGK1ev0iwWmCTbybbJXL/VcvFygxue5t4HHgCdmR0uOdxqmG3NmG3u0u3uYmdz7NxjutS/zuponpNj7jl3kEIkB3G2WKMwlWK84RitaemEHGBXW2+xpFljcSpSpCXTvW2Wy47R2lnKwWmMG4kmTHZDsknV+mgmlFK/aFAWYsK3c3wzI3YNoW0JTUPqOlLoIHeYPKdZ7EJu0MpjjWI4GEJWcoRNwszLSZOCkrlT0kRpOgHRDEpAz2qMEaTgZS2hOiriTIfvJhzs3cZ3DVU9xNkKhRClMSLwNtphbIWrR1CNqTeOs376FMfvuptqfJ4vXQj85z+4xaWLC7pg6Ry87JGHeMePP8rrH3kTRdF3hChS0hzu7PFLP/c+fvMXfpFmZ4oGxsdLvvMffStv/wd/j8GJMd4mAh6dI2a1QNC9wHh1HfFs4NBRiPjROdSgUsbRUeYDZptP8vQTv89s9xpkjzYKV2iKMjMYZIoioLW8RjlqQpdomkTbaEJU5CxzVG3EJ51SInhoFtAsUp8jrAhdR7M8pGsnnDh1mmqwRsqGmOXYNqrufX4XvUuP/8tHYw9qFMsOvdVGMENGK6yBshDaqlICXIwpCRVXZUJOfeHLJJPA9no6l9FOQAVFYXBWYXoQpDOaQmsKa1Am9QoRRUiw9IaJt0zSkFzfwemzD3LXi17FsbsfRK2dJhRDljETomQXaCUas6absVgccniwzcHWDte+8GWe/qtPMb16nbFqOTHM3Hu24syJEb5LzJdJLvYkeKH93R0e/+u/5vLlK9x9772cPPUcZJU2HD95iq96wyPcc/99fPnSZXZ2tmUZgCCYDqeBixcnbN9KHBufxOTMfGeLyc1b+O0JRZspQkL7gM2anNRRsI+1Dmc1RU/eSDH2RzOR6IxGlpOna4oyCrDUKujlMq6QcCa9sq+FDtop051NJrsTTpy4m2pwEnRFRpZH8t4WaFWis0PnAqsryHJTUznSzueE+YK4XNIuZ3TdnORbUpiT8xJNYDmbk3wi+EBRGspa3D2xS1LwvCYHTQ5y1CNB9LEvboILo/dm56yIMaIIGCWZLU4lShsxumU+3WG6v4N1jsFghFYFGQu6AO3QRjh01lkhEQ82sIM7uXrd8psf+hu+8NkdvHcEB/e98qW88z3/goff+BbKwqJ7+GtMlr3dPX7pA/+OX/+5D+B3J7gcqerAC1+ywd/9tke45wXnidbQ4vvHGlEpEHOUSaCRLi/FlXbm2aPtyqnS939YAiZNmO1+kYtf+EumWxfQqUNrqEeG4chQ1ZHRWFMWfYaJl6OqjHtgupAl25EIPYmGUWWDxhKDIvqE0wVkiD5hVMJHj0+ZeryGLYdkU/SSled5p/fUE+97NCDHn5wzRvVpXhmUAeck0UwryEnElQKPTAKo1Aja2lqMNdhSo2zGOBkGmz7cRSuZWVmncUaKn1EKaxUhJ9qkWSbLIpXM84jy+H2cf9FDnHnw1YzuvB8GY2IxpNNWjt5J0q1i8rRtS4jS4W1vb7J5/Srt3gGHl28zuXyT9ZQYm8h6CRsjzx1nEnffPcKgmU4zrZeBrwZS2/HMF57mzz/yEQajAXedP0dV1Ud36LKqeOGLXswb3vImZs2Sy9eu0LRL6KknzdJz9eouF57ZJLYK3XXo5VScFD5io2w3U0qytOmH3krL3X+V7EZOWKMlttJmxmuOqoakGpSVsO6kxHDuzHM2d1mWBQqFzR3zvZssJvsM1tew5QhlB6AcOWesMpRGQokKO8SaAdaWGGOpigK8p5ke0M726JpDQjcjxxk5NeTYUFlDMw0000zXgFYS/+iMEK271tO1GaKBJBtGp+X4Kk4Ujc6G0EVB1PfeUxHaiq1RAos0TmcKE4hpzv7OLXJKjMYbGDOQ0YAx2FJ4cU5bTHToVHLj4oxfev8f8/jHv4xvDdEp7n/1K3jXT/0rHnnL11KUtqdda0Jf8D74Cz/PB3/2Z/G7U6qcWR/DQ69d5yX3ayY7VxjWjo21dazT5NxJ0VuFHCnbw08VqD66kXQkfRFhY+p1dBnDDNpbXP3Sx7j2xSfo5hO5zhSUpWSQGOepaqHWkDXBa0KwdMExnWcOZ5nW02dgFD1eQfdxAkpUA00kR4PvoG2CRGt2ntligqtL1k6cAlORlXv+F72Yn3g0u4pF19G0HTklnMoYLWmaKSEhwTkfDb7FZ6l67Z4cj1Z6MNW7E1eRg0VpUEbcBgKgFPU+KmN1IuTM1CtmyRGrY4zufJA7XvBqTt/3SqqNu4h2jMcRtKNDkVRGpQApkFIg+MB82tA2HbPDGYdbe0xv77B/6SqTq9cYhsB6YRjUmdFIMagVdeHZ2Ajcc0/N6ZM17TIyXQjCSXYsmcn+Pn/9V5/g1u1b3HHmTk6cPIkxcsHqftb38JveyKmzd3Lj9m22d3dIMRwJZCcHDdev7dLNFMeGG6wNLc60GKLoypIE8KQeyxRDFuhojEfe0tjPg6xRKC1MQtOPD4xVxJx6nZwi9+K5lJKIXLPGdx58y2KyxeHudcpSMxiuoe2ATCEeWhROFWgqVG/+t9ZhlSY2M5aTTZbT2zSzHXw7Icem/9OiU0YHw/SgpWsUID5b68A5OSX4LtE1K+inYKsKZ9A5oVNGRXEtxJh61wJoJZGfSUl4kUATDFbBoFJY7Zke7jGfTBgON6iKkYhzjYAWnBpgYsmXntrk1/7Dn/D4x5+maxXZWR589cv5kXe9mzd9zdeJ+P6owzMc7O/zGx/8JX7t5z7AYnOfIiXWxomHHhrylkcG3HvWUqoJmzeeZmvrKsvFDt3yNuQlikCpgoRa5dwLhVW/aZXPPjrIv5fEemZYosM2W9ee4MaFx6Ddx+mMsYaiKHDOSGSpjZSlEFdkwZRomsxkFjmYJDpvCVHhfZTXVfW+7d7k4btEs8yEzpBjIdIzo7FOoBCtXzIYjRiMjoEqGVYPPr+L3sQ/8ehw4xjD9Q2UdqTUE2ozovPpxaQ5q97DCCAXqY/yoU4JQsp9yIx0gtpojE29H5CjnFUjgz1RvilHpwfE6jj1qXs4de/LWTv7IHp0JxRrdKogKEPSmgD4KNmfOkVyjHgf6ZpM03gWsyWLgxnz7V1mNzc5uHiF2ncMbaJwSaxug0g58JQlDGvF+nrgnnsNL3zROsrA3sSzbET0abLCL5Z88anP8cTjj1EPau46dxdVXR/NZKp6wEte9jJe/VXPRVZ5gRfkTNdGdvaXHB50VNUa6yfWKOpEURqsEu2YpPalI6mBCIfBqD6boke259jfSPosWFtInKMMzOUik717r4nzUlx10ujY0c522L75DLGbUJaOoqjRpiZnB1Qyg1IJckcMS1JYcrB7mc2bTxPafdrllNg1xG5BCktUjNDbzIIPdEuP01AWmrJSVLXCGNFximUNYs/mI2XEL8CRmFkr3bPjZIEk80ARjmSV+rGKQiWwKuOMYTbZ5WB3i7IcMBquicXOVJBqLl/Y50O/8id88mOfl39bW+57+Uv44Xf9KF/9dd+Aq9xRwUtJMzk45Nd/+Rf45fe9n/mtHaqcGQ0Tr3ltzSNfZTi51jCulxw7FlkbR8gHTHYucLD1NFuXP8XtC5/jcH8Lv9gnNQfo0FDoLFubDEY5OWoqSMqg8Zi4w8HtJ7nw1EdoD29QO+kIjarRlISU6XxLWWmqSuJVQ5doFolmqTicZLx3uHKEMQW+iSQfnvVza8jZEINhOYfpFFKsgFLQ+06kZYvllNl8wnhtnWqwRl2/8vld9G4f/tmjWRuMqajqNerBGsaWhKgIXjSxIcrdIvV5CiuWmazntWCFnJBZjZKji7Eyj1K63wpGje8UKRtCtiRdoapjlMfPMTp9H4MT92JGp4huRIcjaYPM9YUE3IVIiP0QNwW6pqVZiqZwPl9KNu3uPs3OLtNrN7DzJWOdGRSRapCohxII7QZQr8F4XTEcZIZ1x+kzgQdetMbGiZrpPDCZRmIQ2YiKid3NbT75sb9ic/M2d50/x7Hjx49ApdZa7rjzDA+/8U2cPnOGGzdvsru7fwQvSClzsOi4tTMnxIL1jQ3W1ysqm1AhQAcq9l2zEv6dQRh4SgmyySjTU2RkJpZSFpx8llCgqi76OZ3uxw9ihVOx1+qrTGkSOsyY7l5m//YzLKa7VPUIMzhBVA7vG0KYcXhwg72dZzjYu8D04BqL6abM8WIghZbsW/lQ+Ej2mRQVRlniQja0kClrSz00FKWEq7seAtp5GawTnw1fr8vqWQ2b0hBF+yniaQVY+X9if1oQk7xTgUIpfJPY3dnBFobx+DRarXP5y/v8x//w+3zio0/RLDPZltzz0hfyw+/653ztN7yVqq6lG0IWLZODCb/1Hz/Ir7z//Uxv7uBSZlgnHnrtBm95pObUuGFgOpztcHWkXlNsrGtOjGFctpRmgY4T5vvX2Lr5DLeufJata59lsnOF2zcus3njGsfqNQbFGuiKpDOJCXF2kRtf/HMWu5coaChsQquC5VwzmUTaNqJN6lMHXR9An1gsMtMZNK3BuAHOOclOSQrfhl7kD3IbtMRgmc8ys6licqBol3JSqAegbYciEv2STGJ4/CTjted5Gtrt/T95NHWJ0EZClMSqGKSDm89a9vfmhC4j5A4xdAtayGKNhFPrfkCts8JqQ+EshZNlSEwQgiIETQiGxSJxOEsoO8QOT6IGJ4jFOtGOCKkgZkvScixA9cdhQDykBpUFdhlCpm08zbKVTu9gSrN3wOzWJs3WDgOVGNeJjWOOcpSoxpnRhmEwslRDhRt4yipQV4myCAzHDefvdbzggRPkrNneaWk66fhcUnTzJZ9/6ik+/Tefoqwr7r73HoqiPJr11cMBL3nFy3nZq17FbLng6tWrlv9JcQAAIABJREFU+Nb3eHpousTWzoL9fc+wXmNjfUTtMoSW1CVy6HMglARZryIujer9J0phdP8eZI3vIikJwNMajbVGVPVKbkbZ07fXqV+UaKxOFLTYeMhi9zK7N6/gSstgVNP5Ofv715lMbzBf3mIxv0m73EblVpwXMR4d3fARfCJ2GQKYaMlR0zQeHwLaaIrSUg0c2kaUjbiyB1pohfeZ6DMpyIZXcnaVUFq8FL0VTCBl5O9FEfmGIPaPHCM2W5zOxDjj9tYWKa2zt2X4tX//h3z0z56kWWaSKbj7pS/kHe/+Ub7ubd9IXQ+PyNApw2wy47d//df49z/zXiY3d7A5MxwlXve6db7mTRUn1uaYOEcFKbpeJ5JJGBMoi0hRdZSDxHCkGQ0zG6PAuO6o9QHt/ArTg8sc7j7N/vYl5rM96oFlUGt02mPr6mNsXXyMWnlKZ2Semy1NA5PDlpQyVW1whfAgUxAa93SeWbSKlB0pgStEhN42QRwkvX3TGEtOls4bFvNMs9BMDwM+BAYDuSkpFanKCqMis8WEYrDO6bNvf34Xvc3d3380dS2xkyNjbDqWkzmTnUMOtvdpZq3MiFB47wlBDN0q6SNTvtarAbTYxcii0wox03aJrpPtUoyW0Cnm80DrNVHX4EaoYgimJKJgpedjdTQQmrAYySV82odECpngO3zbCAJpssAfHDLf2sKGjo1BZjTK1EME/FlnBmNDUQeKKlINNGWtsS6jTMYViaJqOXEq8MCDGxzfGHKw52lmwgfUACGwefMmH//YR9nZ3eHeF7yAtfV1EYEqQZOfOXcXj7z5zZy44zTPXLzIdDKVGSYQExxMO27emtEtNWtraxSFuFgIEZ0l8Wsl9M5HYdYZiH1H1MM947NJZkc+TGSmqlMh+r7eA2v7RUdG5j3OJIq8wO9d5/bFp5jv3yT5A/xiE5UOSHlG9HN0aklhiVEZnYDelaIz5JBIPqOiBHtrbfC+z7P1st03VgARykBRQlnBeFxQ1hafMk2X6JpMSiKijRE4ojWLnEbWAL3mrxOxM8mQYylCcbXE6ogOIz7/2Qm/9aHH+cTHn6FtFck4zj74Av77H3snb/3mb6Kshj1hRtLf5tMDPvyhD/Hz730vhze20DkzWlM89PoTfN1bKk6tz1BxKl2pNuhSge2BHBoRVVuFNpGizLgq4cqIsx1VGRlWnrU1z/qww7DLwcGXOJxcZFAsUfNtbn7+cfx0ixhSP/JJxBjQRjEclpw8NaasNGjHcqnxHTRtZNEmshJUlTaJwdBhTKbr2h6RreU6TYauLVgsJCelKgs2NkYUZcaWEaU6tO4Fzl0kd57YBO55+Q88v4ve7q3fe9SEAJ2nnS9YHh4w291jf3OLbrakNFrcF1ofcdNCSKScSEmglCh6emsCnYkp0HlP5yNtl4hJjrVGO1Ci22qaSNNFCbM2FpTBGdnS6ZxQWTaQOfUXspLwl5gyKURi8LTtkq5b0i0XpPmCeDhlsbvDWhnYWIPxmpICVyvqoaaqoSwjg0FmMABTaJKxqEKjXO4pxB11Nef8+ZJ7zx0jNIndfU/bSSHWOdEu53z2b/6GTz3xBOO1Ne46f56iKOVDqBT1cMjLX/UqXvna13AwnXDt5k26zq8smCy7yO3dBXv7HUU5wtVGkuii73NlV2b5fnmk5HlrVgiljDMSASmEYHkvdM9S8z7StbIQUWpFgctYayhLKzNBpzE6wXLK5PZldq99kdn2JfzsNjotKJXHxI7sW1QI5CBdr+q9rSobVI8UC0FIymTBavlG0XXCklNZ5DFVZbBFwBSBaqgZrhUknQlKxhdNj6kiKnJQJK+IQRE6oeyQIXYQGghdXyTxMgNNI3a2a/7w/9rlsSduM2813hrOPnAvP/Jj/4xv/NZvpaq/suAt5hN+5zd+kw/8m/+VveubmBypR4Zv/PbX8Pf/y5cyLDbR8UCO8WgoQNnYO8ENKhuiT0cI+5ijfH5MQvfie3SksJm1yrAxhvG6R5k9ZnsXmd++SjfZx3cNWhf4NksqpQkYFynrjHaeZddyMMns7jYc7HuM0ZSVxZaK4biQiIUCISZXAhBVSmat8yncutUxX0R8yljXUtUBV3q07dAmk6Oja6UrTDHgsufe173j+V30tq99+NHYetrZgmY2Ybm3w8HmbRYHB1gCo5HkcWYSPnlijiI7qTSSOZNBJ5KKZB1QOotzwwFWEzN0SSBHxloBOWromsCy6YipBw+oPsxHadlAgXSSsi8VF0Hf8amcCb6l6eaE2KK6FtN45ts7DJhz+oTj+HHHcJSpa8Vo7CgrqGsJ+alLochEm1CFIhtHMhbtlDx+7zGh4dhG4p771qjrgu29lnkjxV6lJF3f9Zt84uMfZzKZcM999zFeWzuyFhlrOHvuPG9485sZr69x8dIlprNZv5mFkGBv1nF7e0nbasrhEFsocvZCP4iyldWyGBc0qHkWIQRILetvCsln6Y7QsuHzUbJh+7t+SiIIt1aCmqwxUgRSxuaEi548nxDn+yz2btDOd0h+is0B1clxNsVICC0pRJR3EgeQtRxXo1CTyYq2jXSLjG8izmqcMULlKRXaBbLt0EWkHGoGa4pqUOC9BIGnICOU6CUaNCeFbxOp06TWkL0mh0wbvJyyuzHXbwz5nf9zwif/Zp9ZqwhWced95/nhf/4/8U3f9nbqweArCt5yMeX3P/y7/Oy/ei87V29gsmc4tnzjtz3Md/+Tr+feF26gmbOYTGWrnj2oKDeZaI4se6xuACjpUJUsl0QBngXg2tv9FAlbZKoqUqgFhAW+awghEn1kuWz6jlG6YqUjIYt9TlvNaFQxHlfyhutEUYEpggQGKSEQJRVBBUiZ0IFWlrWNEdVI44qELTOKBmsjxgEp0y4DzbKTLA4XKeuOe179Y8/vonfr6ocf9U1LM5sw2d1kb/M6zcEeBZHxyFINLFlFfGzBREllH0A51JRDQ71WUAyhWtdUY1kUlEOHKRXZZFJ/Hso9yUH38yeFUHabZknr214qURzFGYb+aJb7TbLMtyw6KTmGE4h0GDwlkXZvjzzd5K4TjuNjRVkGbCHZtGWlsC5QVJK/QalJpSI5BUaOfRlLzjIsTstAWibhtZmWY+cHnH/wHCGXHOw3dF3oFwuZdrbgyU99ms88+STHThzn9J13UpSFmLe1Zjga8cpXv4aXv+oV7O7ucev2bbz3rMK7lz6xc+iZzjPWDXGlA+Vl+5a1CIdRWG177RW9tKNnBvYXXOykE8oBukYWMTKC0GIvSiIwN1bcNbqfRa54dAZQITIuDJX1OLWAcEhYHBIXM5yKqNySaSBHcsh0raft2p6tp8hRE6M6Cu+JXSa2vS1KR1zVB6NXCozcGI1LVCPFxvGasi5ou0C3FFZeCJngMylYdCxJHcTOy7ggFoQw4ur1ij/40zmPfXrCrIHOKE7fd57/7n/8Z3zL27+DwXCA1j08IGfm00P+jw//Dj/zE/8z21euY5JnOLR8zde9iu/63q/mznMDmjBDG4f3isXikNgthDcY+uS6/topq1JuzknBkXRLQnqcs6Ljo8e2ZVnq6ZxRiA4v9Ha9sjRsbBjWj1uG6wZXGdAZbaEeaAajzGCYKUrZtpaldHtlqXEGyAGlEoUVuZFRBmc1ZZUZjCKDcWQwTBSldKFFpRgMHYOhZTDQ1LURvqGNVIPMuZf/+PO76F2+8L89upzsM925xeHWdbrJIQMNxzdq6oEF7fGpQdmEqxT1yFLUktCui4QuE6qIlAODqSCqIEgfJN/TOC1CZb2CggaMlQuwKh2pC7Rdiw+NwBCtJVtH1JbY+0IFf616UKYm5UBUHqMDpUnk5ZTZrQscLxs2BpHadQwGisHICeXFZQnkLiGXCu8gGjFkG2WJsd94hkTyhrCEtlHMGsMyD7AbG9z/int58EX3UBUjdvamLBadbCGRwf6Nq9d47BN/zXQ65fw997C2vi44eaWwheP8Pffwhje9keFoxNXr15lMDkVEBYQMh/PA7kFHSAVlXcnjVTKsN8iRJSECXq36+EC9ogaLBzP1ti+SKPFz1OARkKXq4Q4GEdIqoSDTC2Nygi5EjIL1oWVQJkqbqVzC6QanG0hzNAu0jigTSanF+5acAyFGQkyy5Y8GnQ05QruIdAvp3lwFrupZiC5jCo1xFuOShKevGeqxQxtNu0x0bRJHh7foZIkhCFQ2OXIes7NT8acfXfLEZ2ZMGoO3ijvuO89/+4538vbv+G4GoyFK95TTZGjmS37/936Hn/3XP8PWxauY5BmMDF/z9a/jH/zDr+PEqQIfFyy7jvky4oOHuCQtPbnN5B5rJhIChbNO1AlRLIMSUC5jBkF4rRIDZbxwxBs8WtBIcl1ZZYbrMFgXR0s9LBgMNcORZjg21INIUXmK0uNsFHuey9Q11LUDFXGFoh44tBLkirGe+qjgRYYjxWhk+9/pqIcFrszSMZqIUkIDd6Xi7Euf50Xvmad+8dHpzk0Ob18nTqasV5bjJ4aUNWTlSbrDFH1CWaUoayO5piaibUYV9Nh0OeZmlQReIPhHKSx2pbLvaSxKCBRFaakHFSommvmC1jdkDdqV4EqSdiglFilNgdaOqBJRhz5QOmBix3LzCqbZYb0KDIvYt+8KTEbbhHZJjo4OvAXfo41UVn1UY8YnCZ5ZLCOHh5ppMySWJymOnaNYPyuuDRpOnRizNhww3Z8znbVCt0DILfODQz79qU/xxaef5vz5u7nzzjMY2xubtWa8vs6rXvtaHnjRC7l85Qq3bt466vgSmWXI7PdawWowZDhWWBOPUEVaJbTpD7tKEugwipASKWe0MiIJwWJ0gdEOrS2ltVgFWSdxy1hFSiL3WSHCY5KO2jhNVZc4p8V2qCPWBZzzOBcoi4R1CecSVQ1r65bBwOE7j4++FxMDqfcWJ8ViHkQ6Eg3VoKQeWbQJWBckH1mLNztrTzFIDNcLilFJaDPdTMtGN62wV5YYh+wdOj7yySWfemrGtDW0RrF+5hQ/9I538B3f/T3Uo5HM1VCQDaEL/N9//Ie891/+FDeeuYjNkbKEN/+d1/A9/803c9c9x9A2EJLHd4HQRZaLltgGbLR0rSclwaGtrLNC9ZaZKlqRenahUdIN5izvlzFanqNW6CT+6pw1Ocr7oEyiGICrJOdCm4gxCeuyaOkKCZMvSkVRZBEqFxHjEtbKaydC4wja44rIcKipBplqCEUFhetHGxaMkUXZ6uYXe82r1pKIeOeL3/W3WvTs3+Y/BiIMTTESeye41tINiIWmF4Wu3BbPKmDla/X9FQFWPed7q7+y+nnN0YcF1dMnlDg1JBtC8hNSjIKU6gf5z0pWVg8493AE+l8m6nPde4flz1f8QP+71NGPZ/UsaFF+uRzJVhq4mDIxaZwp0Nb1wmx6i1ikqixVZdFKyV08P/vYovfs7uywv7cn7ojnwBtQYtk7ceIkG8eOQQ+RfPaRZkKWraYPWXh4q27hK9+1I5r1Vz7NLJGYqX/Nssz0Vu9RWlXY1bsj7PijX6G+8oWTJUj/i5TqZ4lKjvXRZEySCzl2vS6zx4IRIfYLLnqW3oq2knrRMfm5g8lelKRBpZ6+UyhsTz3JOfWPvd9YZ03nYb5MxCiebQBXlhzbOI6x7tkPIECW1LjDwwMW83n/umS00YzGQ+pBySrmTKkVh7jvgI9oGhKglJ/7qFf/UfLe5V7ULrDQowbvOS9xPnrZWX0kV3+e8z2Raz37uufVNdbDImTO2/8ulZ/zVPuoSy2z36P57+p3yYV3RFSCZzl//19+/a0XvXq0gY4RneBw6xb7szm+jWxs1NiiJCaFb1uCyf0WTWEcZCNpTKoPl9ZKOGS5C6Qkeqoc+9Qsn+mahO8H1YVxaF3il4qDgzkzH9HDIRun7mR88izV6BjaFWitBVKgE1kFkupdhVHCXmKKmKzR45NMr9/AEDFFojJRFP7O9PKZKCZrnSmszGVEJp8kTCUnCiApzVgbqhRpFjPmy5ZFN8Ucu4ON03czDSd46uld/uiPnubqpT18G+WCQhO0plxf563f9i18/w/9IC9++csxRUFSfZeWM4vJhD/43d/jlz/wAb70macwXgKAJLRRMbaaF5wecP9dBSeHS8q2weUCazTahP5jrvsQIZnjEKWQqj7IW0YPoF0n81SgjYCJlE6jjchfjJWFS/TyvFUWAbqyGqIcV2POJKUI0ZKVofXgUyLEgrazLKae+WFDaDPNIuEbS2ghe43JpvdqJ8Zjx/E7DHfepxifDCgbwURi6t05WfI2iCXt0nC4Gdm9Nmc5FU9sUViK0pJNRzYe7ebcNR7z1jePiBk++8ySEA07127w/p/5aYajIV/71rfhSiuuCx0oqoK3fcu3s7+3xy/823/H5PYWeeH5sz/6a+q64Du/+83U4ygINJvINmBtA/qQtt1EpwUKT2H6bGetqOsCbTMpdgJP0KY/tiuhDTkFWsYmGYRBCCgLuV/+5H5hlaOcOpKSSM6cg3D3lCIlL51uyPhlpmsyxjrKWmMoadsWlKIsSkILbZNIMVFUmWLVPfb5utIgCFk7xSCj4SA3k9UN/2/762+96A3qAZbeT4vlIN9gOpkSduaMhpZ64DAoukYSqEIXKEpNVWps4dBZ92DDHqkTEOuTT3Re0DddmwgdEBWFLVHJMpt1zOeBRQA9HrF26izrp84wWN+gqGuyUf0sUPXA0YzRChUVUZo7mjaQuw5jBlCeYfv2NdxaiSoCMXu6GCgLKEoNMaFz7p+lJlklRaE37WtVkK0lhQ5VdvI8lCaHKZNdzdbNIX/1l1f46Ecusb8zI/Wm/oQmOceZ++/jH//QD/Lt3/EdHDt5ErRaCUxIXeDyhWf4lZ//eX73N3+L+d4BJiXJsQVKBafHjhedG3P3ScVATShDh0OjrEgfBJ9k0NqT6WMWlcxLbSEQVu0SxUiRTEO20n0po0gOtMnYQmMdWGf6DlXRRXlvUlSopLC6YN71HGWdCckSQoVzQwKGNgZCm/FzTTfzhFZiGXVWUhBcJKRAChFtMvXQcvK05fgdgWMnHWUVMKZPVEuKFDwKg0kVzdSwfWPOfKeTYq4zWCnQRSU3F/6f9s40Wq/zLM/XO+3hm86gI50jWbJl2ciOPMeO7dgGHBIKhZRVSBeEISQhJISSxSrJghIagjOQFlpKWQVKQuMwOIVFAmFBG2igNiFz4oCxsbHj2ZZlDWc+37D3O/bHuyWX/q/oIuf5p6Ul6Rx9+zz7fZ/nvq9bOXRpEWbCRaM+rzBDLPDQY2NmTvLEAw/yb9/1TrQ23Pb1L6eoDUJ4kkwMF0d812u/j7Zt+fVf/GWa9U02Nxo+/rG/RIrAd3z3TSxcMEQFTyUiW+sNa9trBL+JUiFTXwqVEWq6a274LnNYoCqD6oEoPaaUCCXwIbPuBILUBaDHCIg8VpnNciBQ9DkgXppIxGGtw4fs+EEoUtL5BeXJJPOUYRRt44gh3wRs64mpwNmc++tsopmRw/JEhlEE5zLNp8jn1mAF7UzTznJkQBeid17rvDc9YxSSmhgSoyjQQrLJ88y2thGTbrhZFBRK0roW6wLCgrKQTMISclZstCRAS0UiC2u9z7oq60CgKEyJEhpvI9NJZOokxXDE3PIK/T3LVMM5yrqHMgppDGhNkgpUxoBHBD74nGnhIs56XBMIE4cq59n0Wzzz/Dp75ySDvqQ0jugSRI3UmVArRMSkvBlzKmfXQh5IK+WIwhMLkEFl9PjWPI/eJ/izu/+WJ56cYduOAgx4kU93t3/LN/OGH/kRrn7xizFF8cJ9JkUm2zv8+cf/hA/+yq/w4F/dB9ZiYjyLH2RUSo7sLTl6YY+9PYtxE0yIKGROoCsdZ/lT4iyNV0qkzsRpWQhUFZFFpBwqioEiioTrxhW61IQY8tJD5M8lxOxvjZ132lqB8wqLxCVJ24LUPUxVY8wAFQp80DjnsXGKt5bkciC1UVmiFGNeXkmZQ5IQOS5g34pkYUkwt1hS9SII3/1wSrwVBCsJtmBr1TLbdCQPRimEztc4VUFUHmEiWubFltSRoGfownHZZXNUvTnUJ+BvHxkzcZInH3qEn3vPHUghufX2l6FL3dklYX5xidf+0A8igA/98vsZnz7DzpblT/7o8/RGNd/2/TexMDJsnXme2alHUc0kcxfLDDSIkOeQOuF8g5IRVcj8+3VA1d3pLoaMzhd58y5T/vNoSFbhrcK2kbaJBAdrraXuQ9XP3MkYwIZEROG9QqsyB/zYGUJ5yj7omGEUpmsbjc0AWNdAOwGipihrIoGQLFoFFBGjEzMZiA5mk0QzhbLKPyPanP+ud96bHh2Op0KCzFcnImyGhJuOmU09QugMECCLYWObg4SipAvs0RC7nFUpCCGja0KC6PLJKgmJTHkoPZt6XJAUvR7zyysMlpYp5xao+n10ZUBLkjFIXQB5UJTIkoMos20pt1GFQ+OTwiFR1ZDV02v4WcPSnGTYh1YHnIWizqw+4xJVjBRV9/Ukla98qsnfWxAkWWDViOPH5/nLuyd8+tNbnFoN3QxNEIQkGMX+I0f47te9nu96zWtYWl7Op9JubuK956nHHuN3f+M3+f0P/zc2Tp1GhYgiN7xKSg7Ml1x9ZMShvYkibKHaFuVFdlCo0HEJBUJofMiDZtkNNHWhMZVAFB7dT+heopoD08vEEGFzI5I6Zvz8uc2tJqZMdrGtp2kTE2uwsg+DAXrQp7+4QG80JAFt0xLbgG8C0VtkbCElorA5uyTOkMKijSQArQ+ZrtMX7F0pmVuQ1EMo6kSIeQsaY0Hwhq2NwGQ7kJyjmXiUEJQmjzSEBmkSFBE6t0M3kEVqTVlolHEYM2F+JKnLRX5XKO7/u22mFh756wf59+99D8pIbrztVoqyRnRWxrmFPXz/m96AtZa7fu1OpqtrbKw2fOSuexDK8U9evo+dE48RJ2eQ1qFTvoIq3WUuFzrr3AjIAnQVkWVCVjHneKTsHAohoYSklgaUpA2JiRMEW0NbE21DCGOUKjriccK2FuEyWFYLxbRNjMeC6XgCKTG/IKn6WRsptSEKiKrTIIZI8OB9dlrMpoHTJ7eRWtLra4YDSVlWBOtxyWf/PEU3x9eEYAl2eN5b0HlveqGqSdaRgqKQMqO7o0eGhsmqwzXTbL6XEq26bZQQFEp1WKOsTfLd4UYL0XHg8kxIihzQEjsrVQg5Ib4c1PT3LDJcWqScn0P1hyRT4I1ElAapNJE8jc2ZCvnvl1LgRYffMQXBOZwuidJCUZB0xc44UIQIFnp9RUpgQ6QnNCGBT5EySQqdryfIhDCSIBSOio2dee6/T/Nnn1jnkQdnzMZkRwDgpECPRnzdN76CH3jzm7n+xpuoej3O7URSZLy1zafuvps7f/W/8MAXvkSYTvPpknydHdWayy8Y8eLL5thTz5DNFm7aIoLKCwGdN3oZOKAzSl3TPZyyI1MHRAWmFylHknKgMT2R5TkpbxNC7DblKsNFfUz4xues0xiYTATbaYHqwiuY33cEUVaoSuOlp7UTsFNaGUFBwON8S4iOEFqsneCdy6JbnccjrnFAoDcUzO3RLC0XVL2AMh4fBbEFQoWdSbY3HTtbLclBYTT92qBUxHTNTpWZFk0pEIXKmRPka3rGOIpujuxIapsDF9bc/rILcKnioYfXsT7xyJcf4D+8+3287affzk1fdxtF8ULjGy0s8f1vegPeB373v36Q6eom26uWP7zrU6w9vY+br/KMKkPCIiMIJ/ImvkykwqMKSWEkRS1BB2ShSFJ2xOj8sKQYaZHMAlgv2XEFsr/CBYeuoBIlzzz4ZZhNQQT6Qw1YfFBEa0ipoCpLdGoZDQT9SncJg5bGepI3WOuREobDDG0Yj12WdkWQyTIYaHp1QYgarXJYU7QlTTNDSIdSLo87tM4EI6EQ1b7z3IH+IU56ukRGDbpFiICqJGZkKEIPQ8V4fZp1VIXorggxU1SERKGzEr/LQgAIKYNBC20IyeXox5gISXbyjEBRQ3++YrjUoxwVyFohCklSGqlLktYIpbuBa9a6nF0GkiI6xnOhRVIZdNlHVh5R9yj6Q2bbM8ZjMClhpMYHj/Y5IyI4g3aRGBxRJ0otkFriombKgBNnFrjnnin33LPKyZMuh3IngSfhtGT5yIW8+gdez3e8+tXsP3Cw8yUDJIIPPP3449z1wTv577/3EdZPPI/yHt1tX40U7F+suPHKFV50UUWfLez6BnbLkZxCGEEUvlsMqQ4ekIGuSmkKo1BFRPZAVgE9AN1L1HOacig7R0wgdTKILJuAEBNCKmJH1A3C0IQKO7qQQ1d+M/suuQlHydbWKk2zQbBrBJfwcUaKimA9bdMlo9kpth0TQ0sSEaUNJpXZQ+sDZZlY3FOwb3/F/KIi4XAu0kwksx1BO/ZMdzz4RCE7CVSR0FXM13edMgi5BFGCKDtHQ4dOz9klERFaZKzwsWDsE2uu5Zpbb+SSaw9w569/gr/+0qM4F3jwi1/mP/7sv+Mny3fy4ptvwpgKQb7eLy0f4PU//C8hJe56//uxmxPWTzs++b+ew4/nueWGIYMyItMU7eloOJkurosM+RQ6EoTGOYPzmYbtvcjC+zYytppJC6a8gPm9x7jw6EtY2H8A4hZz0wnTh09SCI+SCWsjSlTMppKN9QalHXMLgv5cYmFBo2TCeUg7iVnrCRhA0po857NNniFL4SmrxGiQMApciIy3PeMtydZGS4qRPUslZiAAhymzkL0VFfNHrjj/Leh8/4NKGlSZKR6pcSQ5I4oxQu0g+pFalcgUidGjuiAfYsz2nBAJPoeoxM4bdVaaoMjZDdoktBS58QVPWSmGwqDLiFFjSlGjZIXQNaKqCYUiCHPOpwt5hhVCzmItVNeAYyAUMpNz44zQ75FGlrR3L25nQrszYTxN2U9YRsqeyFftwlP3EzIopJYEBbJn2Jkt8Dd/p/nE3Wvcf/9EeHaIAAAVT0lEQVSYyTjmiD0RcVKgejUvvf1reeOPvoUbb7mVqu4hzkk/YDad8Jm/+CR3/vKvcu+nPk2azdApaxWRUJeKSw/Ncf2xBQ4vCwq7Rrs+xo5DfjurDFpQXdbv/0nbzYy9mGkldcQMEtWCQdcJVUM5UnngLwUxhSwUFgm6CMAYydihqLCiIKllqpWrOHL5NzI8eCPIEcq2LC7tYXvrecRYY5DMfGLbjvFtS2tntM0YN9uB2KBUQkuNFga8xG3NKAtJfyhZ2FMwHEmEcNgm0EwEO5uJ7Y28BCmVoFdKyhIGowIvLOiEKhRCBURBl7eS85KF6kgtJkHS2feroW0FW7OabVuz/8hlHLj4GEns43Wmxv/nj/HAXz2GtS0Pfv7L/PzPvJd//e6f5robb8QUFdnXLFhaXuH1P/wWnHf83od+A7u2w9ZG5DOf2UCmRW66fshcmdDSoZVHFwJdFRhdkqKmmXmmLtC2ilkrmHmN8wqtS6rBMqFaoByMuPrKr2Vx36WEssbpSEwlS4euYvvM00zOPI7yDVplaZR3EcHZHJqA1oJeTyKlR1mfORwp0tqALiuSyGCHsixx7ZiqiAxqqCqPKSTGa2bTQEweOpKhdx2RRyYaa/GiYm7/JSwe/CpoeiUBRUv0a+ysPUO7cRLdbqH8FJN89ktKQQhkIWWnG/I+4jqKihR5KA6d3Ml7CiEpZey8hyknpRUKrSBGT0o7iOmUmdvGb5+hWFhmuHwhZbGPgELJiiRU1hR1P8BnUeImKlRpUFJgbSLFOqOuhEIEaCcNm5OnkCESpomylTgr8S5Q1RGsJM2yVY6qYmdjjk9+oeEvPrvGiecd0ee9axQJLwULB/fzmje9kX/xPd/DgYOHzuVmJCDFwInnjvMbH/gAf/DhD7P53EmUzwTdREJqyb5RydVH9/KiSyv6epM4bTPqZ8chQkTKvABQSnXElkQ6y3vLLk+kDqi+pFoQlCMo5yCZLI1wWGRKnQsgn4qEybD1GBTORdokmckRS0dewvJlX8dg35WoeoWYapQoqcoCIfv06j7OrRDcNqeeuY/t06cygt61hOCz+FhLlIRSGYpUMN6Y4rGMas3cvKGqgBSYTRyuUWytBrbWAwSBUVBWgrInKMpEkC0osle7TKgqB1AlERAmIQuf55JSZqsXeZnQxpq1WWSWSlYuuZ6lg0cJyhBpuOzqC3jtD/0zPvgrH+PB+54kuJa/+ewX+bmfeQ9vf8+7uPYlNyONRHSMvqWVZd70oz+G9Z6P3Pkh3HbDznbkM59dI9HnpusMC/2MTSMK2qlhY1IybhNtMrggiHHIYO/F7Fs6yGCwh8HcPobz+4mqJqmSohjgpcYLS0wSnfro4UWsXHwDX9naxI5PUadEcJ7+oKLXz1frJMcUhUBpjylShvKmhFSKjW2PC022v5FwfkpZBOZGgrqMVL28LbYEiiJQVZJ6uc7ofulxrkFimHowo32sXPwSev1D57sF/QNsb2dr7Kw/y+pzX6HZfJ4elspAISNeCEInzJWKrOQPWU0uu4CfqBRE2WnfIMdIqpyz0YkqZdf8UsoMOMhyBI3F2YbpZIPN9RNsnDnB3guOMlo6TCEF6IqAzp5dYg6m7sSgWY4hURpCgKLoURQlzkfq1tK2LTvPnSJYAR5IHqMULmqSAxsEtulz4tGKT39hg/setkxa312jBUGA6NW8+Oab+OG3/ituu/1lFGX1wukOaJopX/rcZ/m1//RLfP7ue7rTXScWlYKq1BxeGfLiy5bYv8ci22fxOxbXCoLPaV9aBUSXWSoVxJSZaKmLaosClIGiL6mGAjMA3csnQ5cCWkqSFCSRA9NTAqF0Jj9bjwsBK/rE4UEuvuLlrLzoZYj+QUIq8SFihKOQDlJJRJFkhanyy6nf38PcaJn1yXqOa0SgTJk/X5koRUGcQjMJea1kZL4qyZwQ18wEm+uW2XZEpIzxqvua3kB0boyUZRxGEpUn6UTSAlFIdJFA+/wyEBqJyrGPWjFtYXOmsHofFx69hXr+MA3Z3ZDiDEXgiiuW+d7XvoLf8n/KVx46Ds7xwOe+yC+89728/d3v5dg116GMQpCzmveurPCWt/0EMQY+9lt30WxMiduBz39hQmGG3HDNEKEk6zsNm00ijfYyWDnE4vI8C/suYDS/wrA3QqgBSfS6sKJMO0bkOTJymq+fsQZREdVe9hy8jkOTMU//3WdoJus5uTvN0NplekolkdIhFZiC7hnJbiOXYHPc0tpMoTHGMRhA3SNj+0uBkgnvHEoF+gOJbx1ta/HeQ0ikVKAHS1x87DYW9l9NVHvPdws6/za0ez/+ljsmpx9H2Q36ytHXkUqLDtV9NoH9hatrBoxkI3jIy7QOAZ8V7MqIzMQTZ6PvOJf+JMULf19KHXRUS5SIKJFIdsrO2mnWTx7HN2OqIpNFsuM2z7pSJzOPxKxT0wVF0aeqBwitQQoCkd5wiLewdnIDbCT5CFGThMYqzeqkz733Jf78L3d4+MlAYztVvUgELRkdWObVb3wDb/03P8U119+AMSXnXB0h8Pxzz/LhO+/kF9/zPh65969Q1mE69b0yir37B1x/w0GuOjZkWG/hJ6vQBKSXmZ1HoNBgpKQ0iqJQKJW3hIWRCBJGZ62XMoLBqGQ4VyBMyE1CJNAJXeUcEqUlUikSHW8vClLSNL5HsXQNF934new5ejuxPEBAE2KLlhGtcv4qAiKdn9Zv4+wm060zjDdXaaeT3FCkwGiNKQqMltTG0GzPwAfqEuYXNMM5gxCS6cSzvWVppxEtNWWpqHqKXl/n76FIUICuFFHlU52pVXZimIApQpeLrFFCEmKkjbDdajanA8zoUg5ccitmeIhZBBez8TF6n0nPsWHf/hGHLznI8WfPsHZ6i+QSzx8/ztPPPsVlx46xuLQv/593oNW6P+Do5VewNd7m0a88TGoDoUlsrDuErmmjZLOtOXzVN3DVbd/ORVfdxtyBr2G4eIiit0BUfYIc4kSVk8VEzpGFs26XLMbOriQFIuPWhoMe7XTMxpnTOWrTeVKKne0z/3/UfUFZ5WdEKYnSClUYhEh4H9AqMj+nmJ8rqCtBUUlMkeMEnI85oCkZXJuYTS3eelwUiGLEBZfewMGvuQWKFYLoM6j+kYd9n7r37Xf0VEulI6VMaEFmhYW8lPCdSltKA53xP1uy6Oi3OQsiuExNCT4ryUld+LCWWQ2uVCfIDFn1nejiG7O4WUsoFBgsKkxpds6wevoE480zGDuh1GdTv0QWawJKlRgzoC7nMcUw6wnLGlNWCKlpW8fG2gaTyZQoJE4UWGpOrA343JdnfOnBMWe2Ej4mkog5eaxXcu1tt/Djd/wM3/l9r2Hfyv5z11lSomlm3Pv5z/Lz73o3f/jbv8P4+dOUMW+thZTUA80V1xzi277jWq66dg9ajxnvbNHOIhKNlglFpFY51tpIkWM3tc65BUohJTktrXNzIECXiv58hS4jusz6PFlJRCHQReah0dHego84C9Mwh1y4jiM3fy+jC28iFnPEFIEWIbKWLqSE9w4XpvgwpWm2sHaDne0TrK8+w2S8jvcOH7I7QsqCQmsqo8C3tJNtemWi30vML5ToQjCbeSZjR+wE3toIdJEoawHaEaTtrrKd/q0AXWdghTK+w1/p/MylhE+BqZdsNiVTv5fRvmtZOngt6HmamPAkootE64nWYv2UpD3l0LByeJHl/Xt4+onjbK1OSE7w3LNP89iTj3P02DEWl5a7OWpAiMRgOMfRFx1jc2uDxx55mGQDzSzx3Kkpjax46T99Jdfc9k2YPQdoTZE1pEIjUEQ0UWSpVb6OvwAoEF0ymZASoVRnr1QINFpqCq2ws23sdAMRXaZdF7Lb5mdQqCnoAAoe2eWGyM52WRUw7ENVRozJYUI5w13hrGQ2TQSrM+S1TVljWQ9YOnQlFx59KeXwMEGOiEKd9zS08369XaxjbmyddzMG8NHjI9iYCJ3ZXTnfbVMFAoNMAeE90gl0EMSQr7dSnYWAJqTJH6oPIZviYySkjq0n8vYunA0wkQkpI7UWaBlJyjHxZ5ieOcPTzz9B0VtkzwVHmFs+QDlaoKpHJN0jpRJBiZIVWvcxah6tB/lhEvka/fT9T7B1ZsrGtODM054nnlrn1FaDC1nln2G9gsX9K3zrq76d7/vBN3Lk6GVI9cLHkWLk5Inj/PEf/AG/c+eHOP7woxiXKLNyDiET80uSG7/+EN/0z69ibg+srW4ytSPKVhLVGn66SfQp44l8pqRI8kvG+YD3AWR+WSSRpQdKZcnNzsxTjh0jo0i+ybcnmZ0VPnaRmKIL85YVrR5RrlzPgWOvpNx3jFZpUsopZojUnZQFxBaJwAdHiBYfGuxsk52dM7R+TNIBWWqM76OlRiVPpQNF2GFrfZ26lPQNVKWmqkp2dlp2xjbTgEW2QQkZKAqJKhJRZhiErASqp/K2Vku0jmiV+XtCZF2nJ0MGJo1ibIeY3hH2r1xN0duPx2CjJ+JIIUKb8I3FRY9NlrqvKdyUAsGVRwOv/JaDfGRrixPPtPiZ4t57Pskv9d7Hj/3UOzh25dX538WjleDQRYf50R//SZIPfOKjv0+709BuJx593HFyVeMYYGSJUomYZAaTiZxvloQGNDLGzIAU4dxpL1ceiueENElIBq3m2bNyDOyYyfpzTJsZCom1GRKRlMR6KFKGPxQagpcZ2ZbSOVZgWUbKEkyRG2VMHu8E1mvaVtI0EW8F5Ew++sO9XHL5zcztvRwr5ogY/r5p+fzU+YeI3n/HHS7kxPucSyw7LV3CB7LhussvcC5vbfES6SWpBawgOZVR3g40ChE7GKjMD4T1AecyfTd1H/pZiYvoTkjZ799JAroIStWRZ6UIRDtj68wJNk8dZ7a9Rl0X9AYjtK4xsocWNQUVShp0lcGnEFlcWEabPfzNA6d46JEtHnl6h42xzYsZsu+UwnD1jTfwtne+g+967etYOXQIuuAfAGdb/vreL/IL7/1ZPvqh32T92efQscutEIpeLTh2WcU3f8sS192kWDogkIWh8YIgNFEXlP08JzNKoSWo1DlDYiL6iIt5vpVUvh6bokBIgU/Z7xpJKJ0wJegKRJHQVbYtZDFsDoTxsWKHPZQHbuTCa76VYvEIkxBwYYIQtjOzC7LqsIskFJGYWnyc0tptbLODs1OcnRBc0+V8ZMlPaST9siI1Hrs9phSCYa/AGMF0atnanuFclmAYnSiqhK5S3jSXEVmCriWylEgDqoCyCJRVoiwMWhuSSDgCDsXOrMDGZeZXbmLP/uuR5T4cmiY2hDSD4EmzQJrmMPDxzhrRbSLdGL95mq1nH2Tz2QfoyTH9uuLMact07MFLnnnmKU6cPMFlV1zJ4uI+RDeXFlIwHC3woiuuYm1zjUcf+wreJmbThqeeOsHc/AJHjuSMlCQEUqp8Mk0S/g+mnpDx7/UQIcQ5hP9ZOZEQGiENyIpBb0jyls31M0gi3ltctOhCMpovqXqgTRbmCzKXUskshtcma1Kz/S2DQ4SWpFTQzjSbm57p2BO8ROsaXe/l0qu+gX0XvxSv9xJkmbmLRIb/2E96oduGZRKFIIqOx5AUyXer85ggRmTMVN4YHQUFcQa2yXmmKWXuQ7QBpVMOD1KS1rYkGdE6byWJ2U+b4JyDIcbcDPMbPmahc8qECCMDskx4HYlO4H3D5LlVHtt4luHyJawcvJbB/MWU/XlCEphihE+G0XAPfTnPyWaNhx55mPsfXGdjdZq/p5gJE1EpRnv38q2v+nZe9+Yf4uKjRxGmIFPmsixg7cxp/uijH+W33v8Bjn/lMaQLmG6WGVVkca/m1htHvOz2AYtL20zaNXZONgxWCsqypjfMImtfC9KghNkEuTMmnFrH7kwwlnOEjWQEusyLG+ttDs1WMn8+KiKUJqb8MjIInA+5gaWcdWupsXo/84dvZc/hW5CDQ9gkSMlm0XkyCCAk28UDZrAryRPCjOAnRD8lpfxFaamIQhKNRogSnMSEnKUw3XHMJhEfLVpIhPSMZw5EPmmUCspCoCuykFqnDCtRElVmPL82gsIItIoYna95LjosARsLWjsgqRUW919LNTqMSwU2BgINPsyIjUc0gjCe0WyvMZuskuIU5JgNN4HWYkILocEIw7FLhng7z913b3DylKedRD718f+JkoqfeOe7uPhrjnYosJybcsHhw7z5rW9jOpty9x/+D5hZnnv8OB/61Q+jtOHlr3wZpmeIInZo/tQhtTKrUErIcarpHM0kn/CyY0drjZAZpBFEhSyXufBFX8t0Z5PV4/dTIYhyTNXrbJSq28jHPPNT4uzyKKsN8nU2gzgieR4qZX6J1j1FtCBjj8CIlSPXs/+SmwjFEkGURFxGTYX/m+jz/77O/0zvgXffITyI7puNXQ5GDDmpSqFRSaNShQyG2EJqEtFCO43YBqxNSGEgKZzNWQcpdB+0B5kMWlUoWbyAooqxeyiyvUgKEN3pSSjyoFblVbwkoVWiLCR1oegXUOqGZmeDM88/w8bqE0R/hiQaTK2zFzJUPPXQNnd98I/50499ju3VKcLHjgQlCFpx5MoreOs73sFr3/wmVg4dImVvUZ6MBc+jD/0tv/i+9/Hb7/8Am8efR3dh4FEkdAGXX9bjVa9a5uWvUOxbWqUWE1TMyfFS9amHyyALlEmoUiLrAjPsMbdvicGeObYnO1kWQ8IlQZAZspDTwQJCJVSlKKs8z9EG6r7uBv+CEB2EQAySaRwy0wc5cPTl7D18C7I+SBB1XiB1JF8h5DlMVUp00IRECo7gG5Jvid4SvEX4Fhlsnh1JiUgCnRTCelaffZZTTz+JnWxDtAQcbQYLYozM3LcSqp6kqAVFX2WStk5gQJVQlipvGHXmMiqdX3g2RhpXMpuNUPXlLB64GVHtp00FNjnasE2wDaKFuONoNtbZPPUUGyeewG+vIpod/M4asdkg+Vl+rn2C6FAisLhQUdaG5081uFlEOHj2qSc5efokV197HcO5hXN4dylhbn6Ry664gudOnOCpx54A59nZ2uaJJ55hZf8KFx2+CCFV3pyLFwLL83OWXqCfnY257E6TKWWJ0tlfi44FZQqJKRJnTj6Fna0jpWM0JxjMZYLQ2RmeNuIcwk3KPPtTJmXHihGc3SBmJFfETiOzcaK1inJ0gEuvu51y8QheDLqFSzyHFjvfMz3x/wPfard2a7d263yV/If+AnZrt3Zrt85n7Ta93dqt3fqqqt2mt1u7tVtfVbXb9HZrt3brq6p2m95u7dZufVXVbtPbrd3ara+q+t9+c676Y0C9jAAAAABJRU5ErkJggg==\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Model 1: Feature extraction transfer learning on 1% of the data with data augmentation\n"
      ],
      "metadata": {
        "id": "c8V1-vWZ3CZ-"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Setup input shape and base model, freezing the base model layers\n",
        "input_shape = (224, 224, 3)\n",
        "base_model = tf.keras.applications.EfficientNetB0(include_top=False)\n",
        "base_model.trainable = False\n",
        "\n",
        "# Create input layer\n",
        "inputs = layers.Input(shape=input_shape, name=\"input_layer\")\n",
        "\n",
        "# Add in data augmentation Sequential model as a layer\n",
        "x = data_augmentation(inputs)\n",
        "\n",
        "# Give base_model the inputs (after augmentation) and don't train it\n",
        "x = base_model(x, training=False)\n",
        "\n",
        "# Pool output features of the base model\n",
        "x = layers.GlobalAveragePooling2D(name=\"global_average_pooling_layer\")(x)\n",
        "\n",
        "# Put a dense layer on as the output\n",
        "outputs = layers.Dense(10, activation=\"softmax\", name=\"output_layer\")(x)\n",
        "\n",
        "# Make a model\n",
        "model_1 = keras.Model(inputs, outputs)\n",
        "\n",
        "# Compile the model\n",
        "model_1.compile(loss=\"categorical_crossentropy\",\n",
        "                optimizer=\"Adam\",\n",
        "                metrics=[\"accuracy\"])\n",
        "\n",
        "# Fit the model\n",
        "history_1_percent = model_1.fit(train_data_1_percent,\n",
        "                                epochs=5,\n",
        "                                steps_per_epoch=len(train_data_1_percent),\n",
        "                                validation_data = test_data,\n",
        "                                validation_steps=int(0.25 * len(test_data)),\n",
        "                                callbacks=[create_tensorboard_callback(\"transfer_learning\",\n",
        "                                                                       \"1_percent_data_aug\")])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KukQA6cw5Uow",
        "outputId": "85405a32-fb1c-4906-993f-622aacb8aa2d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "WARNING:tensorflow:Model was constructed with shape (512, 512, 3) for input KerasTensor(type_spec=TensorSpec(shape=(512, 512, 3), dtype=tf.uint8, name='random_flip_input'), name='random_flip_input', description=\"created by layer 'random_flip_input'\"), but it was called on an input with incompatible shape (None, 224, 224, 3).\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Saving TensorBoard log files to: transfer_learning/1_percent_data_aug/20230106-162542\n",
            "Epoch 1/5\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "WARNING:tensorflow:Model was constructed with shape (512, 512, 3) for input KerasTensor(type_spec=TensorSpec(shape=(512, 512, 3), dtype=tf.uint8, name='random_flip_input'), name='random_flip_input', description=\"created by layer 'random_flip_input'\"), but it was called on an input with incompatible shape (None, 224, 224, 3).\n",
            "WARNING:tensorflow:Model was constructed with shape (512, 512, 3) for input KerasTensor(type_spec=TensorSpec(shape=(512, 512, 3), dtype=tf.uint8, name='random_flip_input'), name='random_flip_input', description=\"created by layer 'random_flip_input'\"), but it was called on an input with incompatible shape (None, 224, 224, 3).\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "3/3 [==============================] - ETA: 0s - loss: 2.4334 - accuracy: 0.1286"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "WARNING:tensorflow:Model was constructed with shape (512, 512, 3) for input KerasTensor(type_spec=TensorSpec(shape=(512, 512, 3), dtype=tf.uint8, name='random_flip_input'), name='random_flip_input', description=\"created by layer 'random_flip_input'\"), but it was called on an input with incompatible shape (None, 224, 224, 3).\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r3/3 [==============================] - 11s 2s/step - loss: 2.4334 - accuracy: 0.1286 - val_loss: 2.2265 - val_accuracy: 0.1924\n",
            "Epoch 2/5\n",
            "3/3 [==============================] - 3s 1s/step - loss: 2.1353 - accuracy: 0.3143 - val_loss: 2.1007 - val_accuracy: 0.2780\n",
            "Epoch 3/5\n",
            "3/3 [==============================] - 4s 2s/step - loss: 1.9362 - accuracy: 0.3714 - val_loss: 1.9913 - val_accuracy: 0.3569\n",
            "Epoch 4/5\n",
            "3/3 [==============================] - 4s 2s/step - loss: 1.7489 - accuracy: 0.5571 - val_loss: 1.8660 - val_accuracy: 0.4342\n",
            "Epoch 5/5\n",
            "3/3 [==============================] - 4s 2s/step - loss: 1.5923 - accuracy: 0.6714 - val_loss: 1.7898 - val_accuracy: 0.4819\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Check out a model summary\n",
        "model_1.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "U08VcpTh5h3L",
        "outputId": "b0f5a81b-d20b-4465-c568-0108afff219c"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_layer (InputLayer)    [(None, 224, 224, 3)]     0         \n",
            "                                                                 \n",
            " data_augmentation (Sequenti  (None, None, 3)          0         \n",
            " al)                                                             \n",
            "                                                                 \n",
            " efficientnetb0 (Functional)  (None, None, None, 1280)  4049571  \n",
            "                                                                 \n",
            " global_average_pooling_laye  (None, 1280)             0         \n",
            " r (GlobalAveragePooling2D)                                      \n",
            "                                                                 \n",
            " output_layer (Dense)        (None, 10)                12810     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 4,062,381\n",
            "Trainable params: 12,810\n",
            "Non-trainable params: 4,049,571\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Evaluate on the full test dataset\n",
        "results_1_percent_data_aug = model_1.evaluate(test_data)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CEK-2KpI8lGc",
        "outputId": "9d351d2e-7898-42f6-95d2-2d9a98209180"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "79/79 [==============================] - 7s 78ms/step - loss: 1.7831 - accuracy: 0.4864\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# How do the model with 1% and data augmentation loss curves look?\n",
        "plot_loss_curves(history_1_percent)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "id": "uZdeRDmB8taF",
        "outputId": "46b2ee52-0850-47d6-8f78-b550471709d6"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Model 2: feature extraction transfer learning model with 1 0% of data and data augmentation"
      ],
      "metadata": {
        "id": "ePa7X1Kd9Hvg"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Get 10% of data... (we already downloaded it)\n",
        "\n",
        "train_dir_10_percent = \"10_food_classes_10_percent/train\"\n",
        "test_dir = \"10_food_classes_10_percent/test\""
      ],
      "metadata": {
        "id": "CYqzR_8hiyCl"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Set data inputs\n",
        "import tensorflow as tf\n",
        "IMG_SIZE = (224, 224)\n",
        "train_data_10_percent = tf.keras.preprocessing.image_dataset_from_directory(train_dir_10_percent,\n",
        "                                                                            label_mode=\"categorical\",\n",
        "                                                                            image_size=IMG_SIZE)\n",
        "test_data = tf.keras.preprocessing.image_dataset_from_directory(test_dir,\n",
        "                                                                label_mode=\"categorical\",\n",
        "                                                                image_size=IMG_SIZE)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-nLEpSvljGT6",
        "outputId": "74e98fbc-6808-44d4-f129-22f3a039edb4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Found 750 files belonging to 10 classes.\n",
            "Found 2500 files belonging to 10 classes.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Create model 2 with data augmentation built in\n",
        "from tensorflow.keras import layers\n",
        "from tensorflow.keras.layers.experimental import preprocessing\n",
        "from tensorflow.keras.models import Sequential\n",
        "\n",
        "# Build data augmentation layer\n",
        "data_augmentation = Sequential([\n",
        "    preprocessing.RandomFlip(\"horizontal\"),\n",
        "    preprocessing.RandomHeight(0.2),\n",
        "    preprocessing.RandomZoom(0.2),\n",
        "    preprocessing.RandomWidth(0.2),\n",
        "    preprocessing.RandomRotation(0.2),\n",
        "    # preprocessing.Rescaling(1./255) if using other model's, you might need to rescale\n",
        "], name=\"data_augmentation\")\n",
        "\n",
        "# Setup the input shape to our model\n",
        "input_shape = (224, 224, 3)\n",
        "\n",
        "# Create a frozen base model (also called the backbone)\n",
        "base_model = tf.keras.applications.EfficientNetB0(include_top=False)\n",
        "base_model.trainable = False\n",
        "\n",
        "# Create the inputs and outputs (including the layers in between)\n",
        "inputs = layers.Input(shape=input_shape, name=\"input_layer\")\n",
        "x = data_augmentation(inputs)\n",
        "x = base_model(x, training=False) # pass augmented images to base model but keep it in inference mode (training off), this also insures batchnorm layers don't get updated\n",
        "x = layers.GlobalAveragePooling2D(name=\"global_average_pooling_2D\")(x)\n",
        "outputs = layers.Dense(10, activation=\"softmax\", name=\"output_layer\")(x)\n",
        "model_2 = tf.keras.Model(inputs, outputs)\n",
        "\n",
        "# Compile\n",
        "model_2.compile(loss=\"CategoricalCrossentropy\",\n",
        "                optimizer=\"adam\",\n",
        "                metrics=[\"accuracy\"])\n",
        "model_2.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "eytZobA7jkji",
        "outputId": "f8fa381b-6faf-460c-c1fb-d235014f396e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model_2\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_layer (InputLayer)    [(None, 224, 224, 3)]     0         \n",
            "                                                                 \n",
            " data_augmentation (Sequenti  (None, 224, 224, 3)      0         \n",
            " al)                                                             \n",
            "                                                                 \n",
            " efficientnetb0 (Functional)  (None, None, None, 1280)  4049571  \n",
            "                                                                 \n",
            " global_average_pooling_2D (  (None, 1280)             0         \n",
            " GlobalAveragePooling2D)                                         \n",
            "                                                                 \n",
            " output_layer (Dense)        (None, 10)                12810     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 4,062,381\n",
            "Trainable params: 12,810\n",
            "Non-trainable params: 4,049,571\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Creating a ModelCheckpointCallback\n",
        "\n",
        "The ModelCheckpoint callback intermediately saves our model (the full model or just the weights) during training. This is useful so we can come and start where we left off."
      ],
      "metadata": {
        "id": "J8sIfXkLkcnr"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Set checkpoint path\n",
        "checkpoint_path = \"10_percent_model_checkpoints_weights/checkpoint.ckpt\"\n",
        "\n",
        "# Create a ModelCheckpoint callback that saves the model's weights only\n",
        "checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(filepath=checkpoint_path,\n",
        "                                                         save_weights_only=True,\n",
        "                                                         save_best_only=False,\n",
        "                                                         save_freq=\"epoch\",\n",
        "                                                         verbose=1)"
      ],
      "metadata": {
        "id": "RhLjDxa1mmYS"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Fit model 2 passing inthe ModelCheckpoint callback"
      ],
      "metadata": {
        "id": "-e6EuDSrovWB"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Fit the model saving checkpoints every epoch\n",
        "initial_epochs = 5\n",
        "history_10_percent_data_aug = model_2.fit(train_data_10_percent,\n",
        "                                          epochs=initial_epochs,\n",
        "                                          steps_per_epoch=len(train_data_10_percent),\n",
        "                                          validation_data=test_data,\n",
        "                                          validation_steps=int(0.25 * len(test_data)),\n",
        "                                          callbacks=[create_tensorboard_callback(\"transfer_learning\",\n",
        "                                                                                 \"10_percent_data_aug\"),\n",
        "                                                     checkpoint_callback])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GL4ReZdzoGCK",
        "outputId": "a19be24e-0351-4f9a-f487-d8d698ab73f5"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Saving TensorBoard log files to: transfer_learning/10_percent_data_aug/20230106-162627\n",
            "Epoch 1/5\n",
            "24/24 [==============================] - ETA: 0s - loss: 1.9901 - accuracy: 0.3293\n",
            "Epoch 1: saving model to 10_percent_model_checkpoints_weights/checkpoint.ckpt\n",
            "24/24 [==============================] - 20s 577ms/step - loss: 1.9901 - accuracy: 0.3293 - val_loss: 1.4347 - val_accuracy: 0.6875\n",
            "Epoch 2/5\n",
            "24/24 [==============================] - ETA: 0s - loss: 1.2967 - accuracy: 0.7200\n",
            "Epoch 2: saving model to 10_percent_model_checkpoints_weights/checkpoint.ckpt\n",
            "24/24 [==============================] - 11s 441ms/step - loss: 1.2967 - accuracy: 0.7200 - val_loss: 0.9367 - val_accuracy: 0.8306\n",
            "Epoch 3/5\n",
            "24/24 [==============================] - ETA: 0s - loss: 0.9918 - accuracy: 0.7680\n",
            "Epoch 3: saving model to 10_percent_model_checkpoints_weights/checkpoint.ckpt\n",
            "24/24 [==============================] - 10s 398ms/step - loss: 0.9918 - accuracy: 0.7680 - val_loss: 0.7572 - val_accuracy: 0.8372\n",
            "Epoch 4/5\n",
            "24/24 [==============================] - ETA: 0s - loss: 0.8171 - accuracy: 0.8053\n",
            "Epoch 4: saving model to 10_percent_model_checkpoints_weights/checkpoint.ckpt\n",
            "24/24 [==============================] - 11s 425ms/step - loss: 0.8171 - accuracy: 0.8053 - val_loss: 0.6650 - val_accuracy: 0.8454\n",
            "Epoch 5/5\n",
            "24/24 [==============================] - ETA: 0s - loss: 0.7286 - accuracy: 0.8160\n",
            "Epoch 5: saving model to 10_percent_model_checkpoints_weights/checkpoint.ckpt\n",
            "24/24 [==============================] - 9s 333ms/step - loss: 0.7286 - accuracy: 0.8160 - val_loss: 0.5832 - val_accuracy: 0.8651\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# What were model_0 results?\n",
        "model_0.evaluate(test_data)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zl1EWBr-ooiI",
        "outputId": "6871f9ea-3a81-4bc6-abdb-de2d0bf2e12d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "79/79 [==============================] - 7s 76ms/step - loss: 0.5510 - accuracy: 0.8612\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.5509843826293945, 0.8611999750137329]"
            ]
          },
          "metadata": {},
          "execution_count": 36
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Check model_2 results on all test_data\n",
        "results_10_percent_data_aug = model_2.evaluate(test_data)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "s3wwYamOp_4J",
        "outputId": "1fd0ef48-1948-47d9-80bd-eccf8456a3f7"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "79/79 [==============================] - 7s 76ms/step - loss: 0.6178 - accuracy: 0.8392\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Plot loss curves\n",
        "plot_loss_curves(history_10_percent_data_aug)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "id": "bMUCvD6CqEPz",
        "outputId": "36c2a6a7-36fd-4871-ef6e-b0791288513b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Loading in checkpointed weights\n",
        "\n",
        "Loading in checkpointed weights returns a model to specific checkpoint."
      ],
      "metadata": {
        "id": "ZuCl4fIqqL36"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Load in saved model weights and evaluate model\n",
        "model_2.load_weights(checkpoint_path)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FSZCdMlLqpEa",
        "outputId": "ba908844-509f-47fb-a5f9-5847b3025c52"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.training.tracking.util.CheckpointLoadStatus at 0x7fbda525a0a0>"
            ]
          },
          "metadata": {},
          "execution_count": 39
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Evaluate model_2 with loaded weights\n",
        "loaded_weights_model_results = model_2.evaluate(test_data)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SXSzlQ98q7BL",
        "outputId": "3fa50c2a-f4ee-4cf8-f70e-348208d84dbd"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "79/79 [==============================] - 6s 76ms/step - loss: 0.6178 - accuracy: 0.8392\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# If the results from our previously evaluated model_2 match the loaded weights, everything has worked!\n",
        "results_10_percent_data_aug == loaded_weights_model_results"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dfqg9BnkrC2x",
        "outputId": "9b16cea3-a27c-43e1-a470-04bf74dec444"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "False"
            ]
          },
          "metadata": {},
          "execution_count": 41
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "results_10_percent_data_aug"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZlAQwWQWrMX7",
        "outputId": "ffca6ea4-ff3c-44cf-f2d2-d9b6ac82a518"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.6178246736526489, 0.8392000198364258]"
            ]
          },
          "metadata": {},
          "execution_count": 42
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "loaded_weights_model_results"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1H5LIjvtrRHX",
        "outputId": "adabe0a0-bfe8-41cd-b147-c3ff4ef5900a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.6178246140480042, 0.8392000198364258]"
            ]
          },
          "metadata": {},
          "execution_count": 43
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Check to see if loaded model results are very close to our previous non-loaded model results\n",
        "import numpy as np\n",
        "np.isclose(np.array(results_10_percent_data_aug), np.array(loaded_weights_model_results))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TPVUakAprSZb",
        "outputId": "3e8cd3ba-55e9-4707-daf2-48b5485aff73"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([ True,  True])"
            ]
          },
          "metadata": {},
          "execution_count": 44
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Check the difference between the two results\n",
        "print(np.array(results_10_percent_data_aug) - np.array(loaded_weights_model_results))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5VRjqh01roFE",
        "outputId": "139ddee3-49ae-4891-ce46-b9eba727ccd9"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[5.96046448e-08 0.00000000e+00]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Model 3: Fine-tuning an existing model on 10% of the data\n",
        "\n",
        "> **Note:** Fine-tuning usually works best *after* training a feature extraction model for a few epochs with large amounts of custom data."
      ],
      "metadata": {
        "id": "9dRV84I1rzi7"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Layers in loaded model\n",
        "model_2.layers"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FreXfkyPtDm1",
        "outputId": "40863fe8-4f96-42e3-c855-6d9798e8ddd1"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<keras.engine.input_layer.InputLayer at 0x7fbe000fb6a0>,\n",
              " <keras.engine.sequential.Sequential at 0x7fbe0317a070>,\n",
              " <keras.engine.functional.Functional at 0x7fbdfc60adf0>,\n",
              " <keras.layers.pooling.global_average_pooling2d.GlobalAveragePooling2D at 0x7fbe00153b20>,\n",
              " <keras.layers.core.dense.Dense at 0x7fbe000fb670>]"
            ]
          },
          "metadata": {},
          "execution_count": 46
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Are these layers trainable?\n",
        "for layer in model_2.layers:\n",
        "  print(layer, layer.trainable)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mXQm2XnYtX6v",
        "outputId": "0ce46278-37d1-4d69-c216-4d74ce1cfd39"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "<keras.engine.input_layer.InputLayer object at 0x7fbe000fb6a0> True\n",
            "<keras.engine.sequential.Sequential object at 0x7fbe0317a070> True\n",
            "<keras.engine.functional.Functional object at 0x7fbdfc60adf0> False\n",
            "<keras.layers.pooling.global_average_pooling2d.GlobalAveragePooling2D object at 0x7fbe00153b20> True\n",
            "<keras.layers.core.dense.Dense object at 0x7fbe000fb670> True\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# What layers are there in our base_model, and are they trainable?\n",
        "for i, layer in enumerate(model_2.layers[2].layers):\n",
        "  print(i, layer.name, layer.trainable)"
      ],
      "metadata": {
        "id": "9Xdqe5Yute_r",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "37cfd29f-e1a6-4900-9737-bfbe1c648ded"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0 input_3 False\n",
            "1 rescaling_2 False\n",
            "2 normalization_2 False\n",
            "3 tf.math.truediv_2 False\n",
            "4 stem_conv_pad False\n",
            "5 stem_conv False\n",
            "6 stem_bn False\n",
            "7 stem_activation False\n",
            "8 block1a_dwconv False\n",
            "9 block1a_bn False\n",
            "10 block1a_activation False\n",
            "11 block1a_se_squeeze False\n",
            "12 block1a_se_reshape False\n",
            "13 block1a_se_reduce False\n",
            "14 block1a_se_expand False\n",
            "15 block1a_se_excite False\n",
            "16 block1a_project_conv False\n",
            "17 block1a_project_bn False\n",
            "18 block2a_expand_conv False\n",
            "19 block2a_expand_bn False\n",
            "20 block2a_expand_activation False\n",
            "21 block2a_dwconv_pad False\n",
            "22 block2a_dwconv False\n",
            "23 block2a_bn False\n",
            "24 block2a_activation False\n",
            "25 block2a_se_squeeze False\n",
            "26 block2a_se_reshape False\n",
            "27 block2a_se_reduce False\n",
            "28 block2a_se_expand False\n",
            "29 block2a_se_excite False\n",
            "30 block2a_project_conv False\n",
            "31 block2a_project_bn False\n",
            "32 block2b_expand_conv False\n",
            "33 block2b_expand_bn False\n",
            "34 block2b_expand_activation False\n",
            "35 block2b_dwconv False\n",
            "36 block2b_bn False\n",
            "37 block2b_activation False\n",
            "38 block2b_se_squeeze False\n",
            "39 block2b_se_reshape False\n",
            "40 block2b_se_reduce False\n",
            "41 block2b_se_expand False\n",
            "42 block2b_se_excite False\n",
            "43 block2b_project_conv False\n",
            "44 block2b_project_bn False\n",
            "45 block2b_drop False\n",
            "46 block2b_add False\n",
            "47 block3a_expand_conv False\n",
            "48 block3a_expand_bn False\n",
            "49 block3a_expand_activation False\n",
            "50 block3a_dwconv_pad False\n",
            "51 block3a_dwconv False\n",
            "52 block3a_bn False\n",
            "53 block3a_activation False\n",
            "54 block3a_se_squeeze False\n",
            "55 block3a_se_reshape False\n",
            "56 block3a_se_reduce False\n",
            "57 block3a_se_expand False\n",
            "58 block3a_se_excite False\n",
            "59 block3a_project_conv False\n",
            "60 block3a_project_bn False\n",
            "61 block3b_expand_conv False\n",
            "62 block3b_expand_bn False\n",
            "63 block3b_expand_activation False\n",
            "64 block3b_dwconv False\n",
            "65 block3b_bn False\n",
            "66 block3b_activation False\n",
            "67 block3b_se_squeeze False\n",
            "68 block3b_se_reshape False\n",
            "69 block3b_se_reduce False\n",
            "70 block3b_se_expand False\n",
            "71 block3b_se_excite False\n",
            "72 block3b_project_conv False\n",
            "73 block3b_project_bn False\n",
            "74 block3b_drop False\n",
            "75 block3b_add False\n",
            "76 block4a_expand_conv False\n",
            "77 block4a_expand_bn False\n",
            "78 block4a_expand_activation False\n",
            "79 block4a_dwconv_pad False\n",
            "80 block4a_dwconv False\n",
            "81 block4a_bn False\n",
            "82 block4a_activation False\n",
            "83 block4a_se_squeeze False\n",
            "84 block4a_se_reshape False\n",
            "85 block4a_se_reduce False\n",
            "86 block4a_se_expand False\n",
            "87 block4a_se_excite False\n",
            "88 block4a_project_conv False\n",
            "89 block4a_project_bn False\n",
            "90 block4b_expand_conv False\n",
            "91 block4b_expand_bn False\n",
            "92 block4b_expand_activation False\n",
            "93 block4b_dwconv False\n",
            "94 block4b_bn False\n",
            "95 block4b_activation False\n",
            "96 block4b_se_squeeze False\n",
            "97 block4b_se_reshape False\n",
            "98 block4b_se_reduce False\n",
            "99 block4b_se_expand False\n",
            "100 block4b_se_excite False\n",
            "101 block4b_project_conv False\n",
            "102 block4b_project_bn False\n",
            "103 block4b_drop False\n",
            "104 block4b_add False\n",
            "105 block4c_expand_conv False\n",
            "106 block4c_expand_bn False\n",
            "107 block4c_expand_activation False\n",
            "108 block4c_dwconv False\n",
            "109 block4c_bn False\n",
            "110 block4c_activation False\n",
            "111 block4c_se_squeeze False\n",
            "112 block4c_se_reshape False\n",
            "113 block4c_se_reduce False\n",
            "114 block4c_se_expand False\n",
            "115 block4c_se_excite False\n",
            "116 block4c_project_conv False\n",
            "117 block4c_project_bn False\n",
            "118 block4c_drop False\n",
            "119 block4c_add False\n",
            "120 block5a_expand_conv False\n",
            "121 block5a_expand_bn False\n",
            "122 block5a_expand_activation False\n",
            "123 block5a_dwconv False\n",
            "124 block5a_bn False\n",
            "125 block5a_activation False\n",
            "126 block5a_se_squeeze False\n",
            "127 block5a_se_reshape False\n",
            "128 block5a_se_reduce False\n",
            "129 block5a_se_expand False\n",
            "130 block5a_se_excite False\n",
            "131 block5a_project_conv False\n",
            "132 block5a_project_bn False\n",
            "133 block5b_expand_conv False\n",
            "134 block5b_expand_bn False\n",
            "135 block5b_expand_activation False\n",
            "136 block5b_dwconv False\n",
            "137 block5b_bn False\n",
            "138 block5b_activation False\n",
            "139 block5b_se_squeeze False\n",
            "140 block5b_se_reshape False\n",
            "141 block5b_se_reduce False\n",
            "142 block5b_se_expand False\n",
            "143 block5b_se_excite False\n",
            "144 block5b_project_conv False\n",
            "145 block5b_project_bn False\n",
            "146 block5b_drop False\n",
            "147 block5b_add False\n",
            "148 block5c_expand_conv False\n",
            "149 block5c_expand_bn False\n",
            "150 block5c_expand_activation False\n",
            "151 block5c_dwconv False\n",
            "152 block5c_bn False\n",
            "153 block5c_activation False\n",
            "154 block5c_se_squeeze False\n",
            "155 block5c_se_reshape False\n",
            "156 block5c_se_reduce False\n",
            "157 block5c_se_expand False\n",
            "158 block5c_se_excite False\n",
            "159 block5c_project_conv False\n",
            "160 block5c_project_bn False\n",
            "161 block5c_drop False\n",
            "162 block5c_add False\n",
            "163 block6a_expand_conv False\n",
            "164 block6a_expand_bn False\n",
            "165 block6a_expand_activation False\n",
            "166 block6a_dwconv_pad False\n",
            "167 block6a_dwconv False\n",
            "168 block6a_bn False\n",
            "169 block6a_activation False\n",
            "170 block6a_se_squeeze False\n",
            "171 block6a_se_reshape False\n",
            "172 block6a_se_reduce False\n",
            "173 block6a_se_expand False\n",
            "174 block6a_se_excite False\n",
            "175 block6a_project_conv False\n",
            "176 block6a_project_bn False\n",
            "177 block6b_expand_conv False\n",
            "178 block6b_expand_bn False\n",
            "179 block6b_expand_activation False\n",
            "180 block6b_dwconv False\n",
            "181 block6b_bn False\n",
            "182 block6b_activation False\n",
            "183 block6b_se_squeeze False\n",
            "184 block6b_se_reshape False\n",
            "185 block6b_se_reduce False\n",
            "186 block6b_se_expand False\n",
            "187 block6b_se_excite False\n",
            "188 block6b_project_conv False\n",
            "189 block6b_project_bn False\n",
            "190 block6b_drop False\n",
            "191 block6b_add False\n",
            "192 block6c_expand_conv False\n",
            "193 block6c_expand_bn False\n",
            "194 block6c_expand_activation False\n",
            "195 block6c_dwconv False\n",
            "196 block6c_bn False\n",
            "197 block6c_activation False\n",
            "198 block6c_se_squeeze False\n",
            "199 block6c_se_reshape False\n",
            "200 block6c_se_reduce False\n",
            "201 block6c_se_expand False\n",
            "202 block6c_se_excite False\n",
            "203 block6c_project_conv False\n",
            "204 block6c_project_bn False\n",
            "205 block6c_drop False\n",
            "206 block6c_add False\n",
            "207 block6d_expand_conv False\n",
            "208 block6d_expand_bn False\n",
            "209 block6d_expand_activation False\n",
            "210 block6d_dwconv False\n",
            "211 block6d_bn False\n",
            "212 block6d_activation False\n",
            "213 block6d_se_squeeze False\n",
            "214 block6d_se_reshape False\n",
            "215 block6d_se_reduce False\n",
            "216 block6d_se_expand False\n",
            "217 block6d_se_excite False\n",
            "218 block6d_project_conv False\n",
            "219 block6d_project_bn False\n",
            "220 block6d_drop False\n",
            "221 block6d_add False\n",
            "222 block7a_expand_conv False\n",
            "223 block7a_expand_bn False\n",
            "224 block7a_expand_activation False\n",
            "225 block7a_dwconv False\n",
            "226 block7a_bn False\n",
            "227 block7a_activation False\n",
            "228 block7a_se_squeeze False\n",
            "229 block7a_se_reshape False\n",
            "230 block7a_se_reduce False\n",
            "231 block7a_se_expand False\n",
            "232 block7a_se_excite False\n",
            "233 block7a_project_conv False\n",
            "234 block7a_project_bn False\n",
            "235 top_conv False\n",
            "236 top_bn False\n",
            "237 top_activation False\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# How many trainable variables are in our base model?\n",
        "print(len(model_2.layers[2].trainable_variables))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sSF9ht-zt3TD",
        "outputId": "e4b3a9ca-8f5a-4ff5-c821-aa763ee98db6"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# To begin fine-tuning, let's start by setting the last 10 layers of our base_model.trainable = True\n",
        "base_model.trainable = True\n",
        "\n",
        "# Freeze all layers except for the last 10\n",
        "for layer in base_model.layers[:-10]:\n",
        "  layer.trainable = False\n",
        "\n",
        "# Recompile (we have to recompile our models every time we make a change)\n",
        "model_2.compile(loss=\"CategoricalCrossentropy\",\n",
        "                optimizer=tf.keras.optimizers.Adam(lr=0.0001), # when fine-tuning you typically want to lower the learning rate to 10x\n",
        "                metrics=[\"accuracy\"])\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "exU45xuhuaW-",
        "outputId": "e8999a83-9d77-4561-e133-6d244682a6c6"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.8/dist-packages/keras/optimizers/optimizer_v2/adam.py:110: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead.\n",
            "  super(Adam, self).__init__(name, **kwargs)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "> **Note:** When using fine-tuning ot's best practice to lower your learning rate by some amount. How much? This is a hyperparameter you can tune. But a good rule of thumb is at least 10x (though different sources will claim other values).\n",
        "A good resource for information on this is the ULMFiT paper."
      ],
      "metadata": {
        "id": "ABk8BBLLwXIM"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Check which layers are tunable (trainable)\n",
        "for layer_number, layer in enumerate(model_2.layers[2].layers):\n",
        "  print(layer_number, layer.name, layer.trainable)"
      ],
      "metadata": {
        "id": "pPK_KNnnvsxe"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Now we've unfrozen some of the layers closer to the top, how many trainable variables are there?\n",
        "print(len(model_2.trainable_variables))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "maxhjtRNv5Xz",
        "outputId": "2c74b96d-8861-4534-fb08-a3bd8c107c94"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "12\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Fine tune for another 5 epochs\n",
        "fine_tune_epochs = initial_epochs + 5\n",
        "\n",
        "# Refit the model (same as model_2 except with more trainable layers)\n",
        "history_fine_10_percent_data_aug = model_2.fit(train_data_10_percent,\n",
        "                                               epochs=fine_tune_epochs,\n",
        "                                               validation_data=test_data,\n",
        "                                               validation_steps=int(0.25 * len(test_data)),\n",
        "                                               initial_epoch=history_10_percent_data_aug.epoch[-1], # start training from previous last epoch\n",
        "                                               callbacks=[create_tensorboard_callback(\"transfer_learning\", \"10_percent_fine_tune_last_10\")])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vMjUmwOxxNkJ",
        "outputId": "1aae970c-ce3f-4eee-f0c2-ec4ff79bc71f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Saving TensorBoard log files to: transfer_learning/10_percent_fine_tune_last_10/20230106-162816\n",
            "Epoch 5/10\n",
            "24/24 [==============================] - 20s 516ms/step - loss: 0.6026 - accuracy: 0.8373 - val_loss: 0.4579 - val_accuracy: 0.8750\n",
            "Epoch 6/10\n",
            "24/24 [==============================] - 10s 382ms/step - loss: 0.5148 - accuracy: 0.8360 - val_loss: 0.4157 - val_accuracy: 0.8684\n",
            "Epoch 7/10\n",
            "24/24 [==============================] - 10s 400ms/step - loss: 0.4446 - accuracy: 0.8760 - val_loss: 0.4299 - val_accuracy: 0.8668\n",
            "Epoch 8/10\n",
            "24/24 [==============================] - 9s 329ms/step - loss: 0.3931 - accuracy: 0.8747 - val_loss: 0.4131 - val_accuracy: 0.8602\n",
            "Epoch 9/10\n",
            "24/24 [==============================] - 9s 361ms/step - loss: 0.3202 - accuracy: 0.9040 - val_loss: 0.3933 - val_accuracy: 0.8734\n",
            "Epoch 10/10\n",
            "24/24 [==============================] - 9s 367ms/step - loss: 0.2980 - accuracy: 0.9093 - val_loss: 0.3930 - val_accuracy: 0.8684\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Evaluate the fine-tuned model (model_3 which is actually model_2 fined-tuned for another 5 epochs)\n",
        "results_fine_tune_10_percent = model_2.evaluate(test_data)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Tp29CLvAz12y",
        "outputId": "a45f2e62-de9c-4f19-e836-5961f41dc506"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "79/79 [==============================] - 6s 75ms/step - loss: 0.4113 - accuracy: 0.8640\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Check out the loss curves of our fine-tuned model\n",
        "plot_loss_curves(history_fine_10_percent_data_aug)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "id": "L3nR3KHk0gkq",
        "outputId": "06ba651d-f5be-42e4-e042-5c365d310484"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "The `plot_loss_curves` function works great with models which have only been fit once, however, we want something to compare one series of running `fit()` with another (e.g. before and after fine-tuning)."
      ],
      "metadata": {
        "id": "tOSszRP21NZU"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Let's create a function to compare training histories\n",
        "def compare_histories(original_history, new_history, initial_epochs=5):\n",
        "  \"\"\"\n",
        "  Compares two TensorFlow History objects.\n",
        "  \"\"\"\n",
        "  # Get original history measurements\n",
        "  acc = original_history.history[\"accuracy\"]\n",
        "  loss = original_history.history[\"loss\"]\n",
        "\n",
        "  val_acc = original_history.history[\"val_accuracy\"]\n",
        "  val_loss = original_history.history[\"val_loss\"]\n",
        "  \n",
        "  # Combine original history\n",
        "  total_acc = acc + new_history.history[\"accuracy\"]\n",
        "  total_loss = loss + new_history.history[\"loss\"]\n",
        "\n",
        "  total_val_acc = val_acc + new_history.history[\"val_accuracy\"]\n",
        "  total_val_loss = val_loss + new_history.history[\"val_loss\"]\n",
        "\n",
        "  # Make plots fir accuracy\n",
        "  plt.figure(figsize=(8, 8))\n",
        "  plt.subplot(2, 1, 1)\n",
        "  plt.plot(total_acc, label=\"Training Accuracy\")\n",
        "  plt.plot(total_val_acc, label=\"Val Accuracy\")\n",
        "  plt.plot([initial_epochs-1, initial_epochs-1], plt.ylim(), label=\"Start Fine Tuning\")\n",
        "  plt.legend(loc=\"lower right\")\n",
        "  plt.title(\"Training and Validation Accuracy\")\n",
        "    \n",
        "  # Make plots for loss\n",
        "  plt.figure(figsize=(8, 8))\n",
        "  plt.subplot(2, 1, 1)\n",
        "  plt.plot(total_loss, label=\"Training Loss\")\n",
        "  plt.plot(total_val_loss, label=\"Val Loss\")\n",
        "  plt.plot([initial_epochs-1, initial_epochs-1], plt.ylim(), label=\"Start Fine Tuning\")\n",
        "  plt.legend(loc=\"upper right\")\n",
        "  plt.title(\"Training and Validation Loss\")"
      ],
      "metadata": {
        "id": "e6M325-t0oJC"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "compare_histories(history_10_percent_data_aug,\n",
        "                  history_fine_10_percent_data_aug,\n",
        "                  initial_epochs=5)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 507
        },
        "id": "DUI-8nRz2Aw2",
        "outputId": "d63757b7-f04f-4145-a290-ec732905feee"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x576 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x576 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Model 4: Fine-tuning and existing model on all of the data\n"
      ],
      "metadata": {
        "id": "98KpxT-I2sHA"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Download and unzip 10 classes of Fodd101 data with all images\n",
        "!wget https://storage.googleapis.com/ztm_tf_course/food_vision/10_food_classes_all_data.zip\n",
        "unzip_data(\"10_food_classes_all_data.zip\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "z3pXePRu4C8v",
        "outputId": "95e5ef89-67aa-49e5-e8b1-48db1543b8c5"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "--2023-01-06 16:29:38--  https://storage.googleapis.com/ztm_tf_course/food_vision/10_food_classes_all_data.zip\n",
            "Resolving storage.googleapis.com (storage.googleapis.com)... 74.125.31.128, 74.125.28.128, 173.194.211.128, ...\n",
            "Connecting to storage.googleapis.com (storage.googleapis.com)|74.125.31.128|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 519183241 (495M) [application/zip]\n",
            "Saving to: ‘10_food_classes_all_data.zip’\n",
            "\n",
            "10_food_classes_all 100%[===================>] 495.13M   180MB/s    in 2.8s    \n",
            "\n",
            "2023-01-06 16:29:41 (180 MB/s) - ‘10_food_classes_all_data.zip’ saved [519183241/519183241]\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Setup training and test dir\n",
        "train_dir_all_data = \"10_food_classes_all_data/train\"\n",
        "test_dir = \"10_food_classes_all_data/test\""
      ],
      "metadata": {
        "id": "7XiEMmFH4IgF"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# how many images are we working with now?\n",
        "walk_through_dir(\"10_food_classes_all_data\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AhV53j9_4WZ-",
        "outputId": "06567e48-0e0e-4b3c-c904-56f9b6f9976c"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "There are 2 directories and 0 images in '10_food_classes_all_data'.\n",
            "There are 10 directories and 0 images in '10_food_classes_all_data/test'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/chicken_wings'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/steak'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/hamburger'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/sushi'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/ice_cream'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/grilled_salmon'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/chicken_curry'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/ramen'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/pizza'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/fried_rice'.\n",
            "There are 10 directories and 0 images in '10_food_classes_all_data/train'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/chicken_wings'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/steak'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/hamburger'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/sushi'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/ice_cream'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/grilled_salmon'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/chicken_curry'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/ramen'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/pizza'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/fried_rice'.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Setup data inputs\n",
        "import tensorflow as tf\n",
        "IMG_SIZE = (224, 224)\n",
        "train_data_10_classes_full = tf.keras.preprocessing.image_dataset_from_directory(train_dir_all_data,\n",
        "                                                                                 label_mode=\"categorical\",\n",
        "                                                                                 image_size=IMG_SIZE)\n",
        "test_data = tf.keras.preprocessing.image_dataset_from_directory(test_dir,\n",
        "                                                                label_mode=\"categorical\",\n",
        "                                                                image_size=IMG_SIZE)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fv1uB4AD4bJy",
        "outputId": "c60fe4db-3b1d-41c7-87c8-c4186f33b02a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Found 7500 files belonging to 10 classes.\n",
            "Found 2500 files belonging to 10 classes.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "The test dataset we've loaded in is the same as what we've been using for previous experiments (all experiments have used the same test dataset).\n",
        "\n",
        "Let's verify this..."
      ],
      "metadata": {
        "id": "kZEkBhKW5OkP"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Evaluate model 2 (this is the fine-tuned on 10 percent of data version)\n",
        "model_2.evaluate(test_data)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "X5vr3_E246nC",
        "outputId": "6ce9d30c-d6c4-47dd-8ad9-7dd541ed84d2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "79/79 [==============================] - 7s 77ms/step - loss: 0.4113 - accuracy: 0.8640\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.4113459885120392, 0.8640000224113464]"
            ]
          },
          "metadata": {},
          "execution_count": 62
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "results_fine_tune_10_percent"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XROBWnLv5ccL",
        "outputId": "f4a83129-dd7b-4b26-9be8-782e7a15ca8b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.4113459885120392, 0.8640000224113464]"
            ]
          },
          "metadata": {},
          "execution_count": 63
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "To train a fine-tuning model (model_4) we need to revert model_2 back to its feature extraction weights."
      ],
      "metadata": {
        "id": "iRBz55EB5hzV"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Load weights from checkpoint, that way we can fine-tune from the same stage the 10 percent data model was fine-tuned from\n",
        "model_2.load_weights(checkpoint_path)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "D_NhY4Wm5281",
        "outputId": "7b2cf0b6-87bd-4a35-f2c7-bdb68d5d5405"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.training.tracking.util.CheckpointLoadStatus at 0x7fbda42ccf40>"
            ]
          },
          "metadata": {},
          "execution_count": 64
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Let's evaluate model_2 now\n",
        "model_2.evaluate(test_data)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qzLHFifJ6L4T",
        "outputId": "3cf6bdba-e922-48df-be7c-abeae325ddb1"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "79/79 [==============================] - 6s 76ms/step - loss: 0.6178 - accuracy: 0.8392\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.6178244352340698, 0.8392000198364258]"
            ]
          },
          "metadata": {},
          "execution_count": 65
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Check to see if our model_2 has ben reverted back to feature extraction results\n",
        "results_10_percent_data_aug"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kqoScn9w6Qqo",
        "outputId": "032fcf37-444b-4fdd-bb4f-7c622938fdf2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.6178246736526489, 0.8392000198364258]"
            ]
          },
          "metadata": {},
          "execution_count": 66
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Alright, the previous steps might seem quite confusing, but all we've done is:\n",
        "\n",
        "1. Trained a feature extraction transfer learning model for 5 epochs on 10% of the data with data augmentation (model_2) and we saved the model's weights using `ModelCheckpoint` callback.\n",
        "2. Fine-tuned the same model on the same 10% of the data for a further 5 epochs with the top 10 layers of the base model unfrozen (model_3).\n",
        "3. Saved the results and training logs each time.\n",
        "4. Reloaded the model from step 1 to do the same steps as step 2 except this time we're going to use all of the data (model_4)."
      ],
      "metadata": {
        "id": "BTGtvwm26dhP"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Check which layers are tunable in the whole model\n",
        "for layer_number, layer in enumerate(model_2.layers):\n",
        "  print(layer_number, layer.name, layer.trainable)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Zb5lsCtQ7N9F",
        "outputId": "ea63c349-15cf-4d80-9617-f97474a316a3"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0 input_layer True\n",
            "1 data_augmentation True\n",
            "2 efficientnetb0 True\n",
            "3 global_average_pooling_2D True\n",
            "4 output_layer True\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Let's drill into our base_model (EfficientNetB0) and see what layers are trainable\n",
        "for layer_number, layer in enumerate(model_2.layers[2].layers):\n",
        "  print(layer_number, layer.name, layer.trainable)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5cSSfmhp7hl9",
        "outputId": "df9e44e3-dcf0-44a4-838a-d5b54e85321c"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0 input_3 False\n",
            "1 rescaling_2 False\n",
            "2 normalization_2 False\n",
            "3 tf.math.truediv_2 False\n",
            "4 stem_conv_pad False\n",
            "5 stem_conv False\n",
            "6 stem_bn False\n",
            "7 stem_activation False\n",
            "8 block1a_dwconv False\n",
            "9 block1a_bn False\n",
            "10 block1a_activation False\n",
            "11 block1a_se_squeeze False\n",
            "12 block1a_se_reshape False\n",
            "13 block1a_se_reduce False\n",
            "14 block1a_se_expand False\n",
            "15 block1a_se_excite False\n",
            "16 block1a_project_conv False\n",
            "17 block1a_project_bn False\n",
            "18 block2a_expand_conv False\n",
            "19 block2a_expand_bn False\n",
            "20 block2a_expand_activation False\n",
            "21 block2a_dwconv_pad False\n",
            "22 block2a_dwconv False\n",
            "23 block2a_bn False\n",
            "24 block2a_activation False\n",
            "25 block2a_se_squeeze False\n",
            "26 block2a_se_reshape False\n",
            "27 block2a_se_reduce False\n",
            "28 block2a_se_expand False\n",
            "29 block2a_se_excite False\n",
            "30 block2a_project_conv False\n",
            "31 block2a_project_bn False\n",
            "32 block2b_expand_conv False\n",
            "33 block2b_expand_bn False\n",
            "34 block2b_expand_activation False\n",
            "35 block2b_dwconv False\n",
            "36 block2b_bn False\n",
            "37 block2b_activation False\n",
            "38 block2b_se_squeeze False\n",
            "39 block2b_se_reshape False\n",
            "40 block2b_se_reduce False\n",
            "41 block2b_se_expand False\n",
            "42 block2b_se_excite False\n",
            "43 block2b_project_conv False\n",
            "44 block2b_project_bn False\n",
            "45 block2b_drop False\n",
            "46 block2b_add False\n",
            "47 block3a_expand_conv False\n",
            "48 block3a_expand_bn False\n",
            "49 block3a_expand_activation False\n",
            "50 block3a_dwconv_pad False\n",
            "51 block3a_dwconv False\n",
            "52 block3a_bn False\n",
            "53 block3a_activation False\n",
            "54 block3a_se_squeeze False\n",
            "55 block3a_se_reshape False\n",
            "56 block3a_se_reduce False\n",
            "57 block3a_se_expand False\n",
            "58 block3a_se_excite False\n",
            "59 block3a_project_conv False\n",
            "60 block3a_project_bn False\n",
            "61 block3b_expand_conv False\n",
            "62 block3b_expand_bn False\n",
            "63 block3b_expand_activation False\n",
            "64 block3b_dwconv False\n",
            "65 block3b_bn False\n",
            "66 block3b_activation False\n",
            "67 block3b_se_squeeze False\n",
            "68 block3b_se_reshape False\n",
            "69 block3b_se_reduce False\n",
            "70 block3b_se_expand False\n",
            "71 block3b_se_excite False\n",
            "72 block3b_project_conv False\n",
            "73 block3b_project_bn False\n",
            "74 block3b_drop False\n",
            "75 block3b_add False\n",
            "76 block4a_expand_conv False\n",
            "77 block4a_expand_bn False\n",
            "78 block4a_expand_activation False\n",
            "79 block4a_dwconv_pad False\n",
            "80 block4a_dwconv False\n",
            "81 block4a_bn False\n",
            "82 block4a_activation False\n",
            "83 block4a_se_squeeze False\n",
            "84 block4a_se_reshape False\n",
            "85 block4a_se_reduce False\n",
            "86 block4a_se_expand False\n",
            "87 block4a_se_excite False\n",
            "88 block4a_project_conv False\n",
            "89 block4a_project_bn False\n",
            "90 block4b_expand_conv False\n",
            "91 block4b_expand_bn False\n",
            "92 block4b_expand_activation False\n",
            "93 block4b_dwconv False\n",
            "94 block4b_bn False\n",
            "95 block4b_activation False\n",
            "96 block4b_se_squeeze False\n",
            "97 block4b_se_reshape False\n",
            "98 block4b_se_reduce False\n",
            "99 block4b_se_expand False\n",
            "100 block4b_se_excite False\n",
            "101 block4b_project_conv False\n",
            "102 block4b_project_bn False\n",
            "103 block4b_drop False\n",
            "104 block4b_add False\n",
            "105 block4c_expand_conv False\n",
            "106 block4c_expand_bn False\n",
            "107 block4c_expand_activation False\n",
            "108 block4c_dwconv False\n",
            "109 block4c_bn False\n",
            "110 block4c_activation False\n",
            "111 block4c_se_squeeze False\n",
            "112 block4c_se_reshape False\n",
            "113 block4c_se_reduce False\n",
            "114 block4c_se_expand False\n",
            "115 block4c_se_excite False\n",
            "116 block4c_project_conv False\n",
            "117 block4c_project_bn False\n",
            "118 block4c_drop False\n",
            "119 block4c_add False\n",
            "120 block5a_expand_conv False\n",
            "121 block5a_expand_bn False\n",
            "122 block5a_expand_activation False\n",
            "123 block5a_dwconv False\n",
            "124 block5a_bn False\n",
            "125 block5a_activation False\n",
            "126 block5a_se_squeeze False\n",
            "127 block5a_se_reshape False\n",
            "128 block5a_se_reduce False\n",
            "129 block5a_se_expand False\n",
            "130 block5a_se_excite False\n",
            "131 block5a_project_conv False\n",
            "132 block5a_project_bn False\n",
            "133 block5b_expand_conv False\n",
            "134 block5b_expand_bn False\n",
            "135 block5b_expand_activation False\n",
            "136 block5b_dwconv False\n",
            "137 block5b_bn False\n",
            "138 block5b_activation False\n",
            "139 block5b_se_squeeze False\n",
            "140 block5b_se_reshape False\n",
            "141 block5b_se_reduce False\n",
            "142 block5b_se_expand False\n",
            "143 block5b_se_excite False\n",
            "144 block5b_project_conv False\n",
            "145 block5b_project_bn False\n",
            "146 block5b_drop False\n",
            "147 block5b_add False\n",
            "148 block5c_expand_conv False\n",
            "149 block5c_expand_bn False\n",
            "150 block5c_expand_activation False\n",
            "151 block5c_dwconv False\n",
            "152 block5c_bn False\n",
            "153 block5c_activation False\n",
            "154 block5c_se_squeeze False\n",
            "155 block5c_se_reshape False\n",
            "156 block5c_se_reduce False\n",
            "157 block5c_se_expand False\n",
            "158 block5c_se_excite False\n",
            "159 block5c_project_conv False\n",
            "160 block5c_project_bn False\n",
            "161 block5c_drop False\n",
            "162 block5c_add False\n",
            "163 block6a_expand_conv False\n",
            "164 block6a_expand_bn False\n",
            "165 block6a_expand_activation False\n",
            "166 block6a_dwconv_pad False\n",
            "167 block6a_dwconv False\n",
            "168 block6a_bn False\n",
            "169 block6a_activation False\n",
            "170 block6a_se_squeeze False\n",
            "171 block6a_se_reshape False\n",
            "172 block6a_se_reduce False\n",
            "173 block6a_se_expand False\n",
            "174 block6a_se_excite False\n",
            "175 block6a_project_conv False\n",
            "176 block6a_project_bn False\n",
            "177 block6b_expand_conv False\n",
            "178 block6b_expand_bn False\n",
            "179 block6b_expand_activation False\n",
            "180 block6b_dwconv False\n",
            "181 block6b_bn False\n",
            "182 block6b_activation False\n",
            "183 block6b_se_squeeze False\n",
            "184 block6b_se_reshape False\n",
            "185 block6b_se_reduce False\n",
            "186 block6b_se_expand False\n",
            "187 block6b_se_excite False\n",
            "188 block6b_project_conv False\n",
            "189 block6b_project_bn False\n",
            "190 block6b_drop False\n",
            "191 block6b_add False\n",
            "192 block6c_expand_conv False\n",
            "193 block6c_expand_bn False\n",
            "194 block6c_expand_activation False\n",
            "195 block6c_dwconv False\n",
            "196 block6c_bn False\n",
            "197 block6c_activation False\n",
            "198 block6c_se_squeeze False\n",
            "199 block6c_se_reshape False\n",
            "200 block6c_se_reduce False\n",
            "201 block6c_se_expand False\n",
            "202 block6c_se_excite False\n",
            "203 block6c_project_conv False\n",
            "204 block6c_project_bn False\n",
            "205 block6c_drop False\n",
            "206 block6c_add False\n",
            "207 block6d_expand_conv False\n",
            "208 block6d_expand_bn False\n",
            "209 block6d_expand_activation False\n",
            "210 block6d_dwconv False\n",
            "211 block6d_bn False\n",
            "212 block6d_activation False\n",
            "213 block6d_se_squeeze False\n",
            "214 block6d_se_reshape False\n",
            "215 block6d_se_reduce False\n",
            "216 block6d_se_expand False\n",
            "217 block6d_se_excite False\n",
            "218 block6d_project_conv False\n",
            "219 block6d_project_bn False\n",
            "220 block6d_drop False\n",
            "221 block6d_add False\n",
            "222 block7a_expand_conv False\n",
            "223 block7a_expand_bn False\n",
            "224 block7a_expand_activation False\n",
            "225 block7a_dwconv False\n",
            "226 block7a_bn False\n",
            "227 block7a_activation False\n",
            "228 block7a_se_squeeze True\n",
            "229 block7a_se_reshape True\n",
            "230 block7a_se_reduce True\n",
            "231 block7a_se_expand True\n",
            "232 block7a_se_excite True\n",
            "233 block7a_project_conv True\n",
            "234 block7a_project_bn True\n",
            "235 top_conv True\n",
            "236 top_bn True\n",
            "237 top_activation True\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Compile \n",
        "model_2.compile(loss=\"CategoricalCrossentropy\",\n",
        "                optimizer=tf.keras.optimizers.Adam(lr=0.0001),\n",
        "                metrics=[\"accuracy\"])"
      ],
      "metadata": {
        "id": "HGbY2Msz71zR"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Continue to train and fine-tune the model to our data (100% of training data)\n",
        "fine_tune_epochs = initial_epochs + 5\n",
        "\n",
        "history_fine_10_classes_full = model_2.fit(train_data_10_classes_full,\n",
        "                                           epochs=fine_tune_epochs,\n",
        "                                           validation_data=test_data,\n",
        "                                           validation_steps=int(0.25 * len(test_data)),\n",
        "                                           initial_epoch=history_10_percent_data_aug.epoch[-1],\n",
        "                                           callbacks=[create_tensorboard_callback(\"transfer_learning\",\n",
        "                                                                                  \"full_10_classes_fine_tune_last_10\")])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_1bdLb8q8mSV",
        "outputId": "5159998a-9efe-4c35-f403-f2a0740283ad"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Saving TensorBoard log files to: transfer_learning/full_10_classes_fine_tune_last_10/20230106-163004\n",
            "Epoch 5/10\n",
            "235/235 [==============================] - 68s 261ms/step - loss: 0.6501 - accuracy: 0.7919 - val_loss: 0.3593 - val_accuracy: 0.8882\n",
            "Epoch 6/10\n",
            "235/235 [==============================] - 54s 228ms/step - loss: 0.5354 - accuracy: 0.8263 - val_loss: 0.3262 - val_accuracy: 0.9030\n",
            "Epoch 7/10\n",
            "235/235 [==============================] - 51s 215ms/step - loss: 0.4782 - accuracy: 0.8457 - val_loss: 0.2835 - val_accuracy: 0.9194\n",
            "Epoch 8/10\n",
            "235/235 [==============================] - 48s 204ms/step - loss: 0.4220 - accuracy: 0.8633 - val_loss: 0.2967 - val_accuracy: 0.9112\n",
            "Epoch 9/10\n",
            "235/235 [==============================] - 44s 186ms/step - loss: 0.3946 - accuracy: 0.8679 - val_loss: 0.2401 - val_accuracy: 0.9342\n",
            "Epoch 10/10\n",
            "235/235 [==============================] - 45s 191ms/step - loss: 0.3617 - accuracy: 0.8845 - val_loss: 0.2596 - val_accuracy: 0.9309\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Let's evalaute on all of the test data\n",
        "results_fine_tune_full_data = model_2.evaluate(test_data)\n",
        "results_fine_tune_full_data"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zX_sSu549d9n",
        "outputId": "831aa9a5-6a0d-49e6-a4c8-6d6b30b61a13"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "79/79 [==============================] - 6s 75ms/step - loss: 0.2633 - accuracy: 0.9168\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.26326242089271545, 0.9168000221252441]"
            ]
          },
          "metadata": {},
          "execution_count": 71
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# How did fine-tuning go with more data?\n",
        "compare_histories(original_history=history_10_percent_data_aug,\n",
        "                  new_history=history_fine_10_classes_full,\n",
        "                  initial_epochs=5)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 507
        },
        "id": "zNrmYERk-LZq",
        "outputId": "616eca66-0e00-4afd-d86f-1eda569709fd"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x576 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x576 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Viewing our experiment data on TensorBoard\n",
        "\n",
        "> **Note:** Anything you upload to TensorBoard.dev is going to be public. So if  you have private data, do not upload."
      ],
      "metadata": {
        "id": "iT3hRrQ2-gJa"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# View tensorboard logs of transfer learning modelling experiments (should be ~4 models)\n",
        "# Upload TensorBoard dev records\n",
        "!tensorboard dev upload --logdir ./transfer_learning \\\n",
        "  --name \"Transfer Learning Experiments with 10 Food101 Classes\" \\\n",
        "  --description \"A series of different transfer learning experiments with varying amounts of data and fine-tuning.\" \\\n",
        "  --one_shot # exits the uploader once its finished uploading"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "OynLgrxl8xr1",
        "outputId": "19d6e4a1-3b4e-4ccf-d1f9-130d0e71a47d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            "***** TensorBoard Uploader *****\n",
            "\n",
            "This will upload your TensorBoard logs to https://tensorboard.dev/ from\n",
            "the following directory:\n",
            "\n",
            "./transfer_learning\n",
            "\n",
            "This TensorBoard will be visible to everyone. Do not upload sensitive\n",
            "data.\n",
            "\n",
            "Your use of this service is subject to Google's Terms of Service\n",
            "<https://policies.google.com/terms> and Privacy Policy\n",
            "<https://policies.google.com/privacy>, and TensorBoard.dev's Terms of Service\n",
            "<https://tensorboard.dev/policy/terms/>.\n",
            "\n",
            "This notice will not be shown again while you are logged into the uploader.\n",
            "To log out, run `tensorboard dev auth revoke`.\n",
            "\n",
            "Continue? (yes/NO) yes\n",
            "\n",
            "Please visit this URL to authorize this application: https://accounts.google.com/o/oauth2/auth?response_type=code&client_id=373649185512-8v619h5kft38l4456nm2dj4ubeqsrvh6.apps.googleusercontent.com&redirect_uri=urn%3Aietf%3Awg%3Aoauth%3A2.0%3Aoob&scope=openid+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fuserinfo.email&state=PUR8InaUvfVv6pyoDdSwl78jbBpOp4&prompt=consent&access_type=offline\n",
            "Enter the authorization code: 4/1AWgavddfUywHF7nHI8FfjropdDtUpVs24yDW7onpirO4AiR7UsYvER0Pk_U\n",
            "\n",
            "\n",
            "New experiment created. View your TensorBoard at: https://tensorboard.dev/experiment/kfdfaIx6QeyObOLQvA3dww/\n",
            "\n",
            "\u001b[1m[2023-01-06T16:44:12]\u001b[0m Started scanning logdir.\n",
            "\u001b[1m[2023-01-06T16:44:18]\u001b[0m Total uploaded: 162 scalars, 0 tensors, 5 binary objects (4.2 MB)\n",
            "\u001b[1m[2023-01-06T16:44:18]\u001b[0m Done scanning logdir.\n",
            "\n",
            "\n",
            "Done. View your TensorBoard at https://tensorboard.dev/experiment/kfdfaIx6QeyObOLQvA3dww/\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "My TensorBoard experiments are available at: https://tensorboard.dev/experiment/kfdfaIx6QeyObOLQvA3dww/"
      ],
      "metadata": {
        "id": "d5hV6V4r9pYz"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# View all of your uploaded TensorBoard.dev experiments (public)\n",
        "!tensorboard dev list"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "e1zXMq5q97kb",
        "outputId": "4066469c-0b8e-4461-da8e-15a7a5dd6386"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "https://tensorboard.dev/experiment/kfdfaIx6QeyObOLQvA3dww/\n",
            "\tName                 Transfer Learning Experiments with 10 Food101 Classes\n",
            "\tDescription          A series of different transfer learning experiments with varying amounts of data and fine-tuning.\n",
            "\tId                   kfdfaIx6QeyObOLQvA3dww\n",
            "\tCreated              2023-01-06 16:44:12 (9 minutes ago)\n",
            "\tUpdated              2023-01-06 16:44:18 (9 minutes ago)\n",
            "\tRuns                 10\n",
            "\tTags                 5\n",
            "\tScalars              162\n",
            "\tTensor bytes         0\n",
            "\tBinary object bytes  4362220\n",
            "Total: 1 experiment(s)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# To delete an experiment\n",
        "# !tensorboard dev delete --experiment_id {id do experiment}"
      ],
      "metadata": {
        "id": "UjMtmkS6_nli"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "dHITBsyN_7EI"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}